1-Jul-87 23:16:56-PDT,9496;000000000001 Mail-From: LAWS created at 1-Jul-87 22:51:25 Date: Wed 1 Jul 1987 22:49-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #165 To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 2 Jul 1987 Volume 5 : Issue 165 Today's Topics: Queries - Expert Systems in Marketing & KCL on ISI's, Psychology - $6M Man & Methodology ---------------------------------------------------------------------- Date: 1 Jul 87 19:28:15 GMT From: shire.dec.com!morand@decwrl.dec.com Subject: EXPERT SYSTEMS IN MARKETING I'm working on the definition of an DSS for Product pricing positioning and I'm considering the EXPERT SYSTEMS as a potential answer to my problem. Does any body had an experience or know an application of the expert system to marketing ? I would like to take in consideration : - Product life cycle - Price elasticity parameters - Internal competition - External competition Thanks in advance, Jean-claude MORAND DTN 7 821 4782 or (41 22) 87 47 82 DEC Europe decvax!decwrl!rhea!shire!morand ------------------------------ Date: Wed, 1 Jul 87 10:54:49 edt From: Connie Ramsey Subject: KCL on ISI's Has anybody tried to install the latest (documentation dated July 1986) version of KCL on an ISI? We tried, but found that some code was missing when machine=ISI. If anybody knows anything about this problem, we would appreciate a response. Thank you, Connie Ramsey ramsey@nrl-aic.arpa ------------------------------ Date: Tue, 30 Jun 87 15:57:23 MDT From: Raul Machuca STEWS-ID-T 678-4686 Subject: 6Mil man The six-million dollar man has an explanation which is biological rather than psychological. The center on/off receptors of the eye are arranged in a discrete matrix. An edge gives the greatest signal when the edge passes thru the center of a cell. When there is not enough of a signal the edge cannot be seen. An object moving at a fast rate of speed will be seen by the mind as a sequence of snapshots. These snapshots take place when the edge is lined up with the centers of a group of receptors. I an object is moving at a fast rate of speed the neurons will not recover to take another snapshot until the object has moved a considerable distance. The slow motion still frame technique is simulating on film exactly this process. The brain reacts in the same way as if wewere seeing a quickly moving object and thus the neurons generate the same signals as caused by actually looking at something moving at a fast rate of speed. ------------------------------ Date: 1 Jul 87 06:36:44 GMT From: umix!itivax!chinet!lee@RUTGERS.EDU (Lee Morehead) Reply-to: umix!itivax!chinet!lee@RUTGERS.EDU (Lee Morehead) Subject: Re: Why did $6M man run so slowly? It is interesting to note that in the recent sequel movie to the $6M man, his son could run with speeds measured in the hundreds of mph. While Steve and Jamie retained the slow motion special effect, his son was given the video blur special effect to indicate the several times greater speed of his father. Interesting. -- Lee Morehead ...!ihnp4!chinet!lee "One size fits all." Just who is this "all" person anyway, and why is he wearing my clothes? ------------------------------ Date: Tue, 30 Jun 87 07:18:40 pdt From: norman%ics@sdcsvax.ucsd.edu (Donald A. Norman) Subject: On how AI answers psychological issues A comment on sin in AI, or "Why did the $6M man run so slowly AI researchers seem to like the sin of armchair reasoning. It's a pleasant sin: comfortable, fun, stimulating. And nobody can ever be proven right or wrong. Most scientists, on the other hand, believe that real answers are generated through the collection of data, interpreted by validated theories. The question "why did the $6M man run so slowly" is a case in point, but my answer is also stimulated by the conference on "Foundation of AI" that I just attended (held at MIT, arguing about the several theoretical approaches to the representationa and simulation of intelligence). In AIlist, many folks have let forth their theories. Some are clever, some are interesting. Some are probably right, some are probably wrong. How would one ever know which? Letting forth with opinions is no way to answer a scientific question. At the conference, many of AI's most famed luminaries let forth with their opinions. Psychological phenomena made up and explained faster than the speed of thought. Same observation applies. The only thing worse is when a researcher (in any discipline) becomes a parent. then the theories spin wildly and take the form: my child did the following thing; therefore, all children do it; and therefore here is how the mind works. Same for why the $6M man ran so slowly. If you really want to know why slow motion was used, ASK THE FILM MAKER ! (producer, camerman, editor, director). The film maker selected this method for one of several possible reasons, and armchair reasoning about it will get you nowhere. It might have been to stretch out the film, for budgetary reasons, because they didn't know anything else to do, because they accidentally hit the slow-motion switch once and, once they got started on this direction, all future films had to be consistent, etc. One suspects that filmmakers did not go through the long elaborated reasoning that some of the respondents assumed. Whatever the reason, the best (and perhaps only) way to find out is to ask the people who made the decision. Of course, they themselves may not know, given that much of our actions are not consciously known to us and do not necessarly follow from neat declarative rurles stored in some nice simple memory format (which is why expert systems methodology is fundamentally flawed, but that is another story), but at least the verbally described reasons can give you a starting point. Note that the discussion has confounded several different questions. One question is "why did the film makers chose to use slow motion?" A second question is, given that they made that choice, "Why does the slow motion presentation of speeded motion produce a reasonable efffect on the viewer?" Here the answer can only come about through experimentation. However, for this question, the armchair explanations make more sense and can start out as a plausible set of hypotheses to be examined. A third question has gotten raised in the discusion, which is "during times of stress, or incipient danger, or doing a rapid task when very well skilled, does subjective time pass more slowly?" This is an oft-reported finding. Damn-near impossible to test. (Possible, though: subjective time, for example, changes with body temperature, going faster when body temperature is raised, slower when lowered, and since it is possible to determine that fact experimentally, you should be able to determine the other). The nature of subjective time is most complex, but evidence would have it that filled time passes quite differently than unfilled time, and the expert or person intensly focusssed upon events is apt to attend to details not normally visible, hence filling the time interval with numerous more activity and events, hence changing th perception of time. But before you all bombard the net with lots of anectodes about what it felt like when in you auto accient, or skiing incident or ..., let me remind you that the experience you have DURING the event itself, is quite different from your memory of that experience. The esdperimental research on time perception shows that subjective durations can reverse. ( Events that may be boring to experence -- time passes every so slowly -- may be judged to have taken almost no time at all in future retrospections -- no remembered events. Events with numerous things happening -- so quickly that you didn't have time to respond to most of them -- in retropsect may seem to have taken forever.) The moral is that understanding the human (or animal) mind is most difficult, it is apt to come about only through a combination of experimental study, theoretical modeling, and simulation, and armchair thinking, while fun, is pretty irrelevant to the endeavor. Psychology, the field, can be frustrating to the non-participant. Many tedious experiments. Dumb experiments. An insistence on methodology that borders on the insane. And an apparent inability to answer even the simplest questions. Guilty. But for reason. Thinking about "how the mind works" is fun, but not science, not the way to get to the correct answer. don norman Donald A. Norman Institute for Cognitive Science C-015 University of California, San Diego La Jolla, California 92093 norman@nprdc.arpa {decvax,ucbvax,ihnp4}!sdcsvax!ics!norman norman@sdics.ucsd.edu norman%sdics.ucsd.edu@RELAY.CS.NET ------------------------------ End of AIList Digest ******************** 2-Jul-87 00:13:54-PDT,14084;000000000000 Mail-From: LAWS created at 1-Jul-87 22:56:02 Date: Wed 1 Jul 1987 22:54-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #166 To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 2 Jul 1987 Volume 5 : Issue 166 Today's Topics: Theory - Perception, Policy - Quoting ---------------------------------------------------------------------- Date: 29 Jun 87 22:46:31 GMT From: trwrb!aero!venera.isi.edu!smoliar@ucbvax.Berkeley.EDU (Stephen Smoliar) Subject: Re: The symbol grounding problem.... In article <1194@houdi.UUCP> marty1@houdi.UUCP (M.BRILLIANT) writes: > >I was just looking at a kitchen chair, a brown wooden kitchen >chair against a yellow wall, in side light from a window. Let's >let a machine train its camera on that object. Now either it >has a mechanical array of receptors and processors, like the >layers of cells in a retina, or it does a functionally >equivalent thing with sequential processing. What it has to do >is compare the brightness of neighboring points to find places >where there is contrast, find contrast in contiguous places so >as to form an outline, and find closed outlines to form objects. >There are some subtleties needed to find partly hidden objects, >but I'll just assume they're solved. There may also be an >interpretation of shadow gradations to perceive roundness. > I have been trying to keep my distance from this debate, but I would like to insert a few observations regarding this scenario. In many ways, this paragraph represents the "obvious" approach to perception, assuming that one is dealing with a symbol manipulation system. However, other approaches have been hypothesized. While their viability remains to be demonstrated, it would be fair to say that, in the broad scope of perception in the real world, the same may be said of symbol manipulation systems. Consider the holographic model posed by Karl Pribram in LANGUAGES OF THE BRAIN. As I understand it, this model postulates that memory is a collection of holographic transforms of experienced images. As new images are experienced, the brain is capable of retrieving "best fits" from this memory to form associations. Thus, the chair you see in the above paragraph is recognized as a chair by virtue of the fact that it "fits" other images of chairs you have seen in the past. I'm not sure I buy this, but I'm at least willing to acknowledge it as an alternative to your symbol manipulation scenario. The biggest problem I have has to do with retrieval. As far as I understand, present holographic retrieval works fine as long as you don't have to worry about little things like change of scale, translation, or rotation. If this model is going to work, then the retrieval process is going to have to be more powerful than the current technology allows. The other problem relates to concept acquisition, as was postulated in Brilliant's continuation of the scenario: > >Now the machine has a form. If the form is still unfamiliar, >let it ask, "What's that, Daddy?" Daddy says, "That's a chair." >The machine files that information away. Next time it sees a >similar form it says "Chair, Daddy, chair!" It still has to >learn about upholstered chairs, but give it time. > The difficulty seems to be in what it means to file something away if one's memory is simply one of experiences. Does the memory trace of the chair experience include Daddy's voice saying "chair?" While I'm willing to acknowledge a multi-media memory trace, this seems a bit pat. It reminds me of Skinner's VERBAL BEHAVIOR, in which he claimed that one learned the concept "beautiful" from stimuli of observing people saying "beautiful" in front of beautiful objects. This conjures up a vision of people wandering around the Metropolitan Museum of Art mutttering "beautiful" as they wander from gallery to gallery. Perhaps the difficulty is that the mind really doesn't want to assign a symbol to every experience immediately. Rather, following the model of Holland et. al., it is first necessary to build up some degree of reinforcement which assures that a particular memory trace is actually going to be retrieved relatively frequently (whatever that means). In such a case, then, a symbol becomes a fast-access mechanism for retrieval of that trace (or a collection of common traces). However, this gives rise to at least two questions for which I have no answer: 1. What are the criteria by which it is decided that such a symbol is required for fast-access? 2. Where does the symbol's name come from? 3. How is the symbol actually "bound" to what it retrieves? These would seem to be the sort of questions which might help to tie this debate down to more concrete matters. Brilliant continues: >That brings me to a question: do you really want this machine >to be so Totally Turing that it grows like a human, learns like >a human, and not only learns new objects, but, like a human born >at age zero, learns how to perceive objects? How much of its >abilities do you want to have wired in, and how much learned? > This would appear to be one of the directions in which connectionism is leading. In a recent talk, Sejnowski talked about "training" networks for text-to-speech and backgammon . . . not programming them. On the other hand, at the current level of his experiments, designing the network is as important as training it; training can't begin until one has a suitable architecture of nodes and connections. The big unanswered questions would appear to be: will all of this scale upward? That is, is there ultimately some all-embracing architecture which includes all the mini-architectures examined by connectionist experiments and enough more to accommodate the methodological epiphenomenalism of real life? ------------------------------ Date: 1 Jul 87 16:14:41 GMT From: diamond.bbn.com!aweinste@husc6.harvard.edu (Anders Weinstein) Subject: Re: The symbol grounding problem: Against Rosch & Wittgenstein In article <949@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: > >> There is no reliable, consensual all-or-none categorization performance >> without a set of underlying features? That sounds like a restatement of >> the categorization theorist's credo rather than a thing that is so. > >If not, what is the objective basis for the performance? And how would >you get a device to do it given the same inputs? I think there's some confusion as to whether Harnad's claim is just an empty tautology or a significant empirical claim. To wit: it's clear that we can reliably recognize chairs from sensory input, and we don't do this by magic. Hence, we can perhaps take it as trivially true that there are some "features" of the input that are being detected. If we are taking this line however, we have remember that it doesn't really say *anything* about the operation of the mechanism -- it's just a fancy way of saying we can recognize chairs. On the other hand, it might be taken as a significant claim about the nature of the chair-recognition device, viz., that we can understand its workings as a process of actually parsing the input into a set of features and actually comparing these against what is essentially some logical formula in featurese. This *is* an empirical claim, and it is certainly dubitable: there could be pattern recognition devices (holograms are one speculative suggestion) which cannot be interestingly broken down into feature-detecting parts. Anders Weinstein BBN Labs ------------------------------ Date: 1 Jul 87 22:33:50 GMT From: teknowledge-vaxc!dgordon@unix.sri.com (Dan Gordon) Subject: Re: The symbol grounding problem: Against Rosch & Wittgenstein In article <949@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: > > >dgordon@teknowledge-vaxc.ARPA (Dan Gordon) >of Teknowledge, Inc., Palo Alto CA writes: > >> There is no reliable, consensual all-or-none categorization performance >> without a set of underlying features? That sounds like a restatement of >> the categorization theorist's credo rather than a thing that is so. > >If not, what is the objective basis for the performance? And how would >you get a device to do it given the same inputs? Not a riposte, but some observations: 1) finding an objective basis for a performance and getting a device to do it given the same inputs are two different things. We may be able to find an objective basis for a performance but be unable (for merely contingent reasons, like engineering problems, etc., or for more funda- mental reasons) to get a device to exhibit the same performance. And, I suppose, the converse is true: we may be able to get a device to mimic a performance without understanding the objective basis for the model (chess programs seem to me to fall into this class). 2) There may in fact be categorization performances that a) do not use a set of underlying features; b) have an objective basis which is not feature-driven; and c) can only be simulated (in the strong sense) by a device which likewise does not use features. This is one of the central prongs of Wittgenstein's attack on the positivist approach to language, and although I am not completely convinced by his criticisms, I haven't run across any very convincing rejoinder. Maybe more later, Dan Gordon ------------------------------ Date: 1 Jul 87 14:02:28 GMT From: harwood@cvl.umd.edu (David Harwood) Subject: Re: The symbol grounding problem - please start your own newsgroup In article <950@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: [...replying to M.B. about something...] >................................................ I do not see this >intimate interrelationship -- between names and, on the one hand, the >nonsymbolic representations that pick out the objects they refer to >and, on the other hand, the higher-level symbolic descriptions into >which they enter -- as being perspicuously described as a link between >a pair of autonomous nonsymbolic and symbolic modules. The relationship is >bottom-up and hybrid through and through, with the symbolic component >derivative from, inextricably interdigitated with, and parasitic on the >nonsymbolic. Uh - let me get this straight. This is the conclusion for your most recent posting on "the symbol grounding problem." In the first poorly written sentence you criticize to your bogeyman, saying he ain't "perspicuous." Small wonder - you invent him for purposes of obsurantist controversy; no one else even believes in him so far as I can tell. But wait - there is more. You say your bogeyman - he ain't "perspicuous." (as if you aren't responsible for this) Then you go on with what you consider, apparently, to be a "perspicuous" account of the meaning of "names." So far as I can tell, this sentence is the most full and "perspicuous" accounting yet, confirmed by everything you've written on this subject (which I shall not need quote, since it is fresh on everyone's mind). You say, with inestimatable "perspicuity," concerning your own superior speculations about the meaning of names (which I quote since we have all day, day after day, for this): "The relationship is bottom-up and hybrid through and through, with the symbolic component derivative from, inextricably interdigitated with, and parasitic on the symbolic." A mouthful all right. Interdigitated with something all right. Could you please consider creating your own newsgroup, Mr. Harnad? I don't know what your purpose is, except for self-aggrandizement, but I'm fairly sure your purpose has nothing to do with computer science. There's no discussion of algorithms, computing systems, not even any logical formality in all this bullshit. And if we have to hear about the meaning of names - why couldn't we hear from Saul Kripke, instead of you? Then we might learn something. Why not create your own soapbox? I will never listen or bother. I wouldn't even bother to read BBS, which you apparently edit - with considerable help no doubt, except that you don't write all the articles (as you do here). -David Harwood ------------------------------ Date: Wed, 1 Jul 1987 13:28 EDT From: MINSKY%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: AIList Digest V5 #163 Too much, already. This "symbol grounding" has gotten out of hand. This is a network, not a private journal. ------------------------------ Date: Wed 1 Jul 87 22:02:55-PDT From: Ken Laws Reply-to: AIList-Request@STRIPE.SRI.COM Subject: Policy on Quoting Perhaps the discussion of philosophy/theory/perception would be more palatable -- or even concise and understandable -- if we refrained from quoting each other in the style of the old Phil-Sci list. Quotations are often necessary, of course, but the average reader can follow a discussion without each participant echoing his predecessors. Those few who are really interested in exact wordings can save the relevant back issues; I'll even send copies on request. On the whole, I think that this interchange has been conducted admirably. My hope in making this suggestion is that participants will spend less bandwidth attacking each other's semantics and more of it constructing and presenting their own coherent positions. (It's OK if we don't completely agree on terms such as "analog", as long as each contributor builds a consistent world view that includes his own Humpty-Dumpty variants.) -- Ken ------------------------------ End of AIList Digest ******************** 6-Jul-87 01:01:14-PDT,12836;000000000000 Mail-From: LAWS created at 6-Jul-87 00:47:32 Date: Mon 6 Jul 1987 00:46-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #167 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 6 Jul 1987 Volume 5 : Issue 167 Today's Topics: Seminars - Planning Actions with Context-Dependent Effects (SRI) & Automated Process Planning using Abstraction (CMU), Conference - SLUG '87 Reminder & Simulation and AI & Expert Systems in the ADP Environment ---------------------------------------------------------------------- Date: Tue, 30 Jun 87 11:52:11 PDT From: Amy Lansky Subject: Seminar - Planning Actions with Context-Dependent Effects (SRI) SYNTHESIZING PLANS THAT CONTAIN ACTIONS WITH CONTEXT-DEPENDENT EFFECTS Edwin P.D. Pednault (VAX135!EPDP@UCBVAX.BERKELEY.EDU) Knowledge Systems Research Department AT&T Bell Laboratories Crawfords Corner Road Holmdel, NJ 07733 11:00 AM, MONDAY, July 6 SRI International, Building E, Room EJ228 In this talk, I will present an approach to solving planning problems that involve actions whose effects depend on the state of the world at the time the actions are performed. To solve such problems, the idea of a secondary precondition is introduced. A secondary precondition for an action is a condition that must be true at the time the action is performed for the action to have its desired effect. By imposing the appropriate secondary precondition as an additional precondition to an action, we can coerce that action to preserve a desired condition or to cause a desired condition to become true. I will demonstrate the use of secondary preconditions and show how they can be derived from the specification of a planning problem in a completely general and domain-independent fashion. VISITORS: Please arrive 5 minutes early so that you can be escorted up from the E-building receptionist's desk. Thanks! ------------------------------ Date: 30 Jun 87 12:21:51 EDT From: Marcella.Zaragoza@isl1.ri.cmu.edu Subject: Seminar - Automated Process Planning using Abstraction (CMU) SPECIAL SEMINAR TOPIC: AUTOMATED PROCESS PLANNING USING HIERARCHICAL ABSTRACTION * WHO: Dana S. Nau Computer Science Department and Institute for Advanced Computer Studies, University of Maryland, and Factory Automation Systems Division, National Bureau of Standards WHEN: Monday, July 6, 10:00-11:30 a.m. WHERE: WeH 4623 ABSTRACT SIPS is a system which uses AI techniques to decide what machining operations to use in the creation of metal parts. SIPS generates its plans completely from scratch, using the specification of the part to be produced and knowledge about the intrinsic capabilities of each manufacturing operation. Rather than using a rule-based approach to knowledge representation, SIPS uses a hierarchical abstraction technique called hierarchical knowledge clustering. Problem-solving knowledge is organized in a taxonomic hierarchy using frames, and problem solving is done using an adaptation of Branch and Bound. The development of SIPS was done with two long-term goals in mind: the use of AI techniques to develop a practical generative process planning system, and the investigation of fundamental AI issues in representing and reasoning about three-dimensional objects. SIPS represents an important step toward these goals, and a number of extensions and enhancements to SIPS are either underway or planned. SIPS is currently being integrated into the Automated Manufacturing Research Facility (AMRF) project at the National Bureau of Standards. * This work has been supported in part by the following sources: an NSF Presidential Young Investigator Award to Dana Nau, NSF Grant NSFD CDR-85-00108 to the University of Maryland Systems Research Center, IBM Research, General Motors Research Laboratories, and Martin Marietta Laboratories. ------------------------------ Date: Sat, 27 Jun 1987 14:13 CDT From: CS.PURVIS@R20.UTEXAS.EDU Subject: Conference - SLUG '87 Reminder This is a reminder that the national meeting of the Symbolics Lisp Users Group will be held in Seattle, July 6-10th. You may register in advance by calling the University of Washington at (206) 543-2300. The conference schedule is listed below. Note particularly the panel discussions on Thursday and Friday that will examine available alternatives to the Symbolics Lisp development environment architecture and consider what trade-offs are involved. This is THE Lisp machine conference. Don't miss it! SLUG '87 Schedule July 6-10, 1987 - Seattle, Washington MONDAY -- (tutorials) 8:00 Registration desk opens 9:00 to 12:30 * AI Program Design * Overview of Site Administration * Color Graphics I 2:00 to 5:30 * AI Program Design (cont'd) * Overview of Site Administration (cont'd) * Color Graphics II * Color Graphics III TUESDAY -- (tutorials) 8:00 Registration desk opens 9:00 to 12:30 * Programming Productivity I * Introduction to ART * Building Knowledge System Interfaces 2:00 to 5:30 * Programming Productivity II * Introduction to ART (cont'd) 7:00 - 9:00 Reception WEDNESDAY -- (conference sessions) 8:00 Registration desk opens 9:00 to 12:30 * Welcome & Opening remarks * State of SLUG * Symbolics Corporate Status Report * Software & Hardware Support * Technical Status Report * New Product Announcements * General and Reverse Q & A 2:00 to 6:00 * Software Engineering on LISP Machines * Symbolic Computing for New Users * General Technical Q & A Evening -- BOAF (Birds Of A Feather) * Critique of the Symbolics User Interface -- GNU EMACS and HP's NMODE both present a novel way of interacting with LISP. Is the LISP machine paradigm better? This meeting will drive tomorrow afternoon's session. * New user training: Sharing insights, techniques, and introductory materials for new users. * Symbolics maintenance issues. THURSDAY -- (conference sessions) 9:00 to 12:30 * Common LISP -- What is the status of Common LISP the Language? Classes? Common Windows? Error handling? * SLUG Library -- What's new and available? * Networks -- VMS, UNIX, DECNET, IP-TCP, Namespaces, Domain Resolution, etc. * Non-LISP Language Support -- PROLOG, ADA, FORTRAN, PASCAL, C, etc. 2:00 to 5:30 * LISPM pearls -- An informal presentation of useful but little known LISP machine features and capabilities. * Critique of the Symbolics User Interface -- See yesterday's BOAF. * Technical Q & A FRIDAY -- (conference sessions) 9:00 to 12:30 * Trade-offs in LISP (development) environments -- This is a panel discussion of the differences between developing LISP software on different workstation architectures. * Conference Summary & Feedback * SLUG Business Meeting 2:00 to 3:30 * Expert Systems Session ------------------------------ Date: Thu, 25 Jun 87 10:50:35 edt From: Paul Fishwick Subject: Conference - SIMULATION AND AI ANNOUNCEMENT AND CALL FOR PAPERS SIMULATION AND ARTIFICIAL INTELLIGENCE CONFERENCE Part of the 1988 SCS MultiConference San Diego, CA Feb 3-5, 1988 Paper and Special Session Proposals should be sent to SCS (Society for Computer Simulation) by July 15, 1987 [note: the deadline has been extended]. Some suggested topics are listed below: Relation between AI and Simulation Intelligent Simulation Environments Knowledge-Based Simulation Decision Support Systems Qualitative Simulation (there will be a panel discussion on this topic) Simulation in AI Ada and AI and Simulation Aerospace Applications Biomedical Applications Expert Systems in Emergency Planning Automatic Model Generation Expert Systems Learning Systems Natural Language Processing Robotics Speech Recognition Vision AI Hardware/Workstations AI Programming Languages AI/ES Software Tools A paper proposal should be submitted (approx. 300 words) to: SCS P.O. Box 17900 San Diego, CA 92117-7900 ------------------------------------------------------------------------------ People attending the AI and Simulation workshop at AAAI and others interested in AI and Simulation are strongly encouraged to attend! Paul Fishwick University of Florida CSNET: fishwick@ufl.edu ------------------------------ Date: 24 Jun 87 15:35:00 EST From: "LFA" Reply-to: "LFA" Subject: Conference - Expert Systems in the ADP Environment CALL FOR PAPERS ==== === ====== NARDAC Washington/ORNL/DSRD Conference on Expert Systems Technology in the ADP Environment to be held in Washington, D.C. November 2-3, 1987 THE CONFERENCE === ========== The Naval Regional Data Automation Center in Washington, D.C., the Oak Ridge National Laboratory, and the Data Systems Research and Development Program, Martin Marietta Energy Systems, Inc., are sponsoring a conference whose primary focus is on the use of Artificial Intelligence in traditional computing domains and its potential for further exploitation. Both invited talks and contributed papers will be given at the conference. INVITED SPEAKERS ======= ======== Several individuals have tentatively accepted invitations to speak at this conference on the various aspects of Artificial Intelligence as it pertains to traditional computing problems. Scheduled speakers and their topic areas include: Prof. James Slagle (Minnesota) - Keynote speaker Prof. Brian Gaines (Calgary) - Intelligent Interfaces for Knowledge-Based Systems Prof. Larry Henschen (Northwestern) - Logic and Databases Dr. Sukhumay Kundu (Louisiana State) - AI in Software Engineering CONTRIBUTED PAPERS =========== ====== In addition to the invited talks, papers are being solicited from researchers in academia, government and industry in the following areas: ADP Project and Systems Management, Knowledge-Based Simulation and Modeling, Intelligent Man-Machine Interfaces, Intelligent Databases, AI in Software Engineering, AI as a Tool for Decision-Making, and Innovative Applications in MIS or Scientific Computing. SUBMISSION DETAILS ========== ======= Authors are asked to submit five (5) copies of their paper, which is to be single-spaced and between five to seven pages in length. Both finished and ongoing research will be considered by the program committee and referees. Authors should adhere to the following submission schedule: August 1, 1987 - Submission Deadline August 15, 1987 - Notification of acceptance September 15, 1987 - Camera-ready copies due Send papers, requests for additional information, and all other correspondence to Lloyd F. Arrowood Program Chairman Oak Ridge National Laboratory Building 4500-North, Mail Stop 207 Oak Ridge, TN 37831 or BITNET: LFA@ORNLSTC ------------------------------ End of AIList Digest ******************** 6-Jul-87 01:02:25-PDT,19407;000000000000 Mail-From: LAWS created at 6-Jul-87 00:51:49 Date: Mon 6 Jul 1987 00:50-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #168 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 6 Jul 1987 Volume 5 : Issue 168 Today's Topics: Policy - Hard Limit on Quotations, Theory - The Symbol Grounding Problem & Against Rosch and Wittgenstein ---------------------------------------------------------------------- Date: Thu 2 Jul 87 09:41:54-PDT From: Ken Laws Subject: Hard Limit on Quotations The "quotation problem" has become so prevalent across all of the Usenet newsgroups that the gateway now rejects any message with more quoted text than new text. If a message is rejected for this reason, I am unlikely to clean it up and resend. As I indicated last week, I think we could get along just fine with more "I say ..." and less "You said ...". Paraphrases are fine and even beneficial, but trying to reestablish the exact context of each comment is not worth the hassle to the general readership. Perhaps some of the hair splitting could be carried on through private mail, with occasional reports to the list on points of agreement and disagreement. Discussions of perception and categorization are appropriate for AIList, but we cannot give unlimited time and attention to any one topic. I've engaged in "interpolated debate" myself, and have enjoyed this characteristic mode of net discussions. I won't object to occasional use, but I do get very tired of seeing the same text quoted in message after message. I used to edit such repetitions out of the digest, but I can't manage it with this traffic volume. Please keep in mind that this is a broadcast channel and that many readers have slow terminals or have to pay intercontinental transmission fees. Words are money. It seems that a consistent philosophy cannot be put forth in less than a full book, or at least a BBS article, and that meaningful rebuttals require similar length. We have been trying to cram this through a linear channel, with swirls of debate spinning off from each paragraph [yes, I know that's a contradiction], and there is no evidence of convergence. Let's try to slow down for a while. I would also recommend that messages be kept to a single topic, even if that means (initially) that a full response to a previous message must be split into parts. Separate discussion of grounding, categorization, perception, etc., would be more palatable than the current indivisible stream. I would like to sort the discussions, if only for ease of meaningful retrieval, but can't do so if they all carry the same subject line and mix of topics. -- Ken ------------------------------ Date: Thu, 2 Jul 87 09:37:21 EDT From: Alex Kass Subject: AIList Digest V5 #163 Can't we bag this damn symbol grounding discussion already? If it *must* continue, how about instituting a symbol grounding news group, and freeing the majority of us poor AILIST readers from the burden of flipping past the symbol grounding stuff every morning. -Alex ARPA: Kass@yale UUCP: {decvax,linus,seismo}!yale!kass BITNET: kass@yalecs US: Alex Kass Yale University Computer Science Department New Haven, CT 06520 ------------------------------ Date: 2 Jul 87 05:19:05 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem smoliar@vaxa.isi.edu (Stephen Smoliar) Information Sciences Institute writes: > Consider the holographic model proposed by Karl Pribram in LANGUAGES > OF THE BRAIN... as an alternative to [M.B. Brilliant's] symbol > manipulation scenario. Besides being unimplemented and hence untested in what they can and can't do, holographic representations seem to inherit the same handicap as all iconic representations: Being unique to each input and blending continuously into one another, how can holograms generate categorization rather than merely similarity gradients (in the hard cases, where obvious natural gaps in the input variation don't solve the problem for you a priori)? What seems necessary is active feature-selection, based on feedback from success and failure in attempts to learn to sort and label correctly, not merely passive filtering based on natural similarities in the input. > [A] difficulty seems to be in what it means to file something away if > one's memory is simply one of experiences. Episodic memory -- rote memory for input experiences -- has the same liability as any purely iconic approach: It can't generate category boundaries where there is significant interconfusability among categories of episodes. > Perhaps the difficulty is that the mind really doesn't want to > assign a symbol to every experience immediately. That's right. Maybe it's *categories* of experience that must first be selectively assigned names, not each raw episode. > Where does the symbol's name come from? How is the symbol actually > "bound" to what it retrieves? That's the categorization problem. > The big unanswered question...[with respect to connectionism] > would appear to be: will [it] all... scale upward? Connectionism is one of the candidates for the feature-learning mechanism. That it's (i) nonsymbolic, that it (ii) learns, and that it (iii) uses the same general statistical algorithm across problem-types (i.e., that it has generality rather than being ad hoc, like pure symbolic AI) are connectionism's plus's. (That it's brainlike is not, nor is it true, on current evidence, nor even relevant at this stage.) But the real question is indeed: How much can it really do (i.e., will it scale up)? -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 2 Jul 87 04:36:37 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem: Against Rosch & Wittgenstein dgordon@teknowledge-vaxc.ARPA (Dan Gordon) of Teknowledge, Inc., Palo Alto CA writes: > finding an objective basis for a performance and getting a device to > do it given the same inputs are two different things. We may be able > to find an objective basis for a performance but be unable...to get a > device to exhibit the same performance. And, I suppose, the converse > is true: we may be able to get a device to mimic a performance without > understanding the objective basis for the model I agree with part of this. J.J. Gibson argued that the objective basis of much of our sensorimotor performance is in stimulus invariants, but this does not explain how we get a device (like ourselves) to find and use those invariants and thereby generate the performance. I also agree that a device (e.g., a connectionist network) may generate a performance without our understanding quite how it does it (apart from the general statistical algorithm it's using, in the case of nets). But the point I am making is neither of these. It concerns whether performance (correct all-or-none categorization) can be generated without an objective basis (in the form of "defining" features) (a) existing and (b) being used by any device that successfully generates the performance. Whether or not we know know what the objective basis is and how it's used is another matter. > There may in fact be categorization performances that a) do not use > a set of underlying features; b) have an objective basis which is not > feature-driven; and c) can only be simulated (in the strong sense) by > a device which likewise does not use features. This is one of the > central prongs of Wittgenstein's attack on the positivist approach to > language, and although I am not completely convinced by his criticisms, > I haven't run across any very convincing rejoinder. Let's say I'm trying to provide the requisite rejoinder (in the special case of all-or-none categorization, which is not unrelated to the problems of language: naming and description). Wittgenstein's arguments were not governed by a thoroughly modern constraint that has arisen from the possibility of computer simulation and cognitive modeling. He was introspecting on what the features defining, say, "games" might be, and he failed to find a necessary and sufficient set, so he said there wasn't one. If he had instead asked: "How, in principle, could a device categorize "games" and "nongames" successfully in every instance?" he would have had to conclude that the inputs must provide an objective basis which the device must find and use. Whether or not the device can introspect and report what the objective basis is is another matter. Another red herring in Wittegenstein's "family resemblance" metaphor was the issue of negative and disjunctive features. Not-F is a perfectly good feature. So is Not-F & Not-G. Which quite naturally yields the disjunctive feature F-or-G. None of this is tautologous. It just shows up a certain arbitrary myopia there has been about what a "feature" is. There's absolutely no reason to restrict "features" to monadic, conjunctive features that subjects can report by introspection. The problem in principle is whether there are any logical (and nonmagical) alternatives to a feature-set sufficient to sort the confusable alternatives correctly. I would argue that -- apart from contrived, gerrymandered cases that no one would want to argue formed the real basis of our ability to categorize -- there are none. Finally, in the special case of categorization, the criterion of "defining" features also turns out to be a red herring. According to my own model, categorization is always provisional and context-dependent (it depends on what's needed to successfully sort the confusable alternatives sampled to date). Hence an exhaustive "definition," good till doomsday and formulated from the God's-eye viewpoint is not at issue, only an approximation that works now, and can be revised and tightened if the context is ever widened by further confusable alternatives that the current feature set would not be able to sort correctly. The conflation of (1) features sufficient to generate the current provisional (but successful) approximation and (2) some nebulous "eternal," ontologically exact "defining" set (which I agree does not exist, and may not even make sense, since categorization is always a relative, "compared-to-what?" matter) has led to a multitude of spurious misunderstandings -- foremost among them being the misconception that our categories are all graded or fuzzy. -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 2 Jul 87 15:51:40 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem On ailist cugini@icst-ecf.arpa writes: > why say that icons, but not categorical representations or symbols > are/must be invertible? Isn't it just a vacuous tautology to claim > that icons are invertible wrt to the information they preserve, but > not wrt the information they lose?... there's information loss (many > to one mapping) at each stage of the game: 1. distal object... > 2. sensory projection... 3. icons... 4. categorical representation... > 5. symbols... do you still claim that the transition between 2 > and 3 is invertible in some strong sense which would not be true of, > say, [1 to 2] or [3 to 4], and if so, what is that sense?... Perhaps > you just want to say that the transition between 2 and 3 is usually > more invertible than the other transitions [i.e., invertibility as a > graded category]? [In keeping with Ken Laws' recommendation about minimizing quotation, I have compressed this query as much as I could to make my reply intelligible.] Iconic representations (IRs) must perform a very different function from categorical representations (IRs) or symbolic representations (SRs). In my model, IRs only subserve relative discrimination, similarity judgment and sensory-sensory and sensory-motor matching. For all of these kinds of task, traces of the sensory projection are needed for purposes of relative comparison and matching. An analog of the sensory projection *in the properties that are discriminable to the organism* is my candidate for the kind of representation that will do the job (i.e., generate the performance). There is no question of preserving in the IR properties that are *not* discriminable to the organism. As has been discussed before, there are two ways that IRs could in principle be invertible (with the discriminable properties of the sensory projection): by remaining structurally 1:1 with it or by going into symbols via A/D and an encryption and decryption transformation in a dedicated (hard-wired) system. I hypothesize that structural copies are much more economical than dedicated symbols for generating discrimination performance (and there is evidence that they are what the nervous system actually uses). But in principle, you can get invertibility and generate successful discrimination performance either way. CRs need not -- indeed cannot -- be invertible with the sensory projection because they must selectively discard all features except those that are sufficient to guide successful categorization performance (i.e., sorting and labeling, identification). Categorical feature-detectors must discard most of the discriminable properties preserved in IRs and selectively preserve only the invariant properties shared by all members of a category that reliably distinguish them from nonmembers. I have indicated, though, that this representation is still nonsymbolic; the IR to CR transformation is many-to-few, but it continues to be invertible in the invariant properties, hence it is really "micro-iconic." It does not invert from the representation to the sensory projection, but from the representation to invariant features of the category. (You can call this invertibility a matter of degree if you like, but I don't think it's very informative. The important difference is functional: What it takes to generate discrimination performance and what it takes to generate categorization performance.) Finally, whatever invertibility SRs have is entirely parasitic on the IRs and CRs in which they are grounded, because the elementary SRs out of which the composite ones are put together are simply the names of the categories that the CRs pick out. That's the whole point of this grounding proposal. I hope this explains what is invertible and why. (I do not understand your question about the "invertibility" of the sensory projection to the distal object, since the locus of that transformation is outside the head and hence cannot be part of the internal representation that cognitive modeling is concerned with.) -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 2 Jul 87 01:19:35 GMT From: ctnews!pyramid!prls!philabs!pwa-b!mmintl!franka@unix.sri.com (Frank Adams) Subject: Re: The symbol grounding problem: Correction re. Approximationism In article <923@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: |In responding to Cugini and Brilliant I misinterpreted a point that |the former had made and the latter reiterated. It's a point that's |come up before: What if the iconic representation -- the one that's |supposed to be invertible -- fails to preserve some objective property |of the sensory projection? For example, what if yellow and blue at the |receptor go into green at the icon? The reply is that an analog |representation is only analog in what it preserves, not in what it fails |to preserve. I'm afraid when I parse this, using the definitions Harnad uses, it comes out as tautologically true of *all* representations. "Analog" means "invertible". The invertible properties of a representation are those properties which it preserves. Is there some strange meaning of "preserve" being used here? Otherwise, I don't see how this statement has any meaning. -- Frank Adams ihnp4!philabs!pwa-b!mmintl!franka Ashton-Tate 52 Oakland Ave North E. Hartford, CT 06108 ------------------------------ Date: 2 Jul 87 01:07:00 GMT From: ctnews!pyramid!prls!philabs!pwa-b!mmintl!franka@unix.sri.com (Frank Adams) Subject: Re: The symbol grounding problem In article <917@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: |Finally, and perhaps most important: In bypassing the problem of |categorization capacity itself -- i.e., the problem of how devices |manage to categorize as correctly and successfully as they do, given |the inputs they have encountered -- in favor of its fine tuning, this |line of research has unhelpfully blurred the distinction between the |following: (a) the many all-or-none categories that are the real burden |for an explanatory theory of categorization (a penguin, after all, be it |ever so atypical a bird, and be it ever so time-consuming for us to judge |that it is indeed a bird, is, after all, indeed a bird, and we know |it, and can say so, with 100% accuracy every time, irrespective of |whether we can successfully introspect what features we are using to |say so) and (b) true "graded" categories such as "big," "intelligent," |etc. Let's face the all-or-none problem before we get fancy... I don't believe there are any truely "all-or-none" categories. There are always, at least potentially, ambiguous cases. There is no "100% accuracy every time", and trying to theorize as though there were is likely to lead to problems. Second, and perhaps more to the point, how do you know that "graded" categories are less fundamental than the other kind? Maybe it's the other way around. Maybe we should try to understand to understand graded categories first, before we get fancy with the other kind. I'm not saying this is the case; but until we actually have an accepted theory of categorization, we won't know what the simplest route is to get there. -- Frank Adams ihnp4!philabs!pwa-b!mmintl!franka Ashton-Tate 52 Oakland Ave North E. Hartford, CT 06108 ------------------------------ End of AIList Digest ******************** 6-Jul-87 01:03:32-PDT,17610;000000000000 Mail-From: LAWS created at 6-Jul-87 01:00:15 Date: Mon 6 Jul 1987 00:59-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #169 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 6 Jul 1987 Volume 5 : Issue 169 Today's Topics: Theory - "Fuzzy" Categories? ---------------------------------------------------------------------- Date: 2 Jul 87 01:44:00 GMT From: ctnews!pyramid!prls!philabs!pwa-b!mmintl!franka@unix.sri.com (Frank Adams) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <936@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: |The question was: Do all-or-none categories (such as "bird") have "defining" |features that can be used to sort members from nonmembers at the level of |accuracy (~100%) with which we sort? However they are coded, I claim that |those features MUST exist in the inputs and must be detected and used by the |categorizer. A penguin is not a bird as a matter of degree, and the features |that reliably assign it to "bird" are not graded. I don't see how this follows. It is quite possible to make all-or-none judgements based on graded features. Thermostats, for example, do it all the time. People do, too. The examples which come to mind as being obviously in this category are all judgements of actions to take based on such features, not of categorization. But then, we don't understand how we categorize. But to take an example of categorizing based on a graded feature. Consider a typical, unadorned, wooden kitchen chair. We have no problem categorizing this as a "chair". Consider the same object, with no back. This is clearly categorized as a "stool", and not a "chair". Now vary the size of the back. With a one inch back, the object is clearly still a "stool"; with a ten inch back, it is clearly a "chair"; somewhere in between is an ambiguous point. I would assert that we *do*, in fact, make "all-or-none" type distinctions based precisely on graded distinctions. We have arbitrary (though vague) cut off points where we make the distinction; and those cut off points are chosen in such a way that ambiguous cases are rare to non-existent in our experience[1]. In short, I see nothing about "all-or-none" categories which is not explainable by arbitrary cutoffs of graded sensory data. --------------- [1] There are some categories where this strategy does not work. Colors are a good example of this -- they vary over all of their range, with no very rare points in it. In this case, we use instead the strategy of large overlapping ranges -- two people may disagree on whether a color should be described as "blue" or "green", but both will accept "blue-green" as a description. The same underlying strategy applies: avoid borderline situations. -- Frank Adams ihnp4!philabs!pwa-b!mmintl!franka Ashton-Tate 52 Oakland Ave North E. Hartford, CT 06108 ------------------------------ Date: 3 Jul 87 12:43:39 GMT From: ihnp4!homxb!houdi!marty1@ucbvax.Berkeley.EDU (M.BRILLIANT) Subject: Re: The symbol grounding problem In article <958@mind.UUCP>, harnad@mind.UUCP (Stevan Harnad) writes: > On ailist cugini@icst-ecf.arpa writes: > > why say that icons, but not categorical representations or symbols > > are/must be invertible? Isn't it just a vacuous tautology to claim > > that icons are invertible wrt to the information they preserve, but > > not wrt the information they lose?... there's information loss (many > > to one mapping) at each stage of the game ... In Harnad's response he does not answer the question "why?" He only repeats the statement with reference to his own model. Harnad probably has either a real problem or a contribution to the solution of one. But when he writes about it, the verbal problems conceal it, because he insists on using symbols that are neither grounded nor consensual. We make no progress unless we learn what his terms mean, and either use them or avoid them. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201)-949-1858 Holmdel, NJ 07733 ihnp4!houdi!marty1 ------------------------------ Date: 3 Jul 87 19:26:40 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In Article 176-8 of comp.cog-eng: franka@mmintl.UUCP (Frank Adams) of Multimate International, E. Hartford, CT.writes: > I don't believe there are any truly "all-or-none" categories. There are > always, at least potentially, ambiguous cases... no "100% accuracy > every time"... how do you know that "graded" categories are less > fundamental than the other kind? On the face of it, this sounds self-contradictory, since you state that you don't believe "the other kind" exists. But let's talk common sense. Most of our object categories are indeed all-or-none, not graded. A penguin is not a bird as a matter of degree. It's a bird, period. And if we're capable of making that judgment reliably and categorically, then there must be something about our transactions with penguins that allows us to do so. In the case of sensory categories, I'm claiming that a sufficient set of sensory features is what allows as to make reliable all-or-none judgments; and in the case of higher-order categories, I claim they are grounded in the sensory ones (and their features). I don't deny that graded categories exist too (e.g., "big," "smart"), but those are not the ones under consideration here. And, yes, I hypothesize that all-or-none categories are more fundamental in the problem of categorization and its underlying mechanisms than graded categories. I also do not deny that regions of uncertainty (and even arbitrariness) -- natural and contrived -- exist, but I do not think that those regions are representative of the mechanisms underlying successful categorization. The book under discussion ("Categorical Perception: The Groundwork of Cognition") is concerned with the problem of how graded sensory continua become segmented into bounded all-or-none categories (e.g., colors, semitones). This is accomplished by establishing upper and lower thresholds for regions of the continuum. These thresholds, I must point out, are FEATURES, and they are detected by feature-detectors. The rest is a matter of grain: If you are speaking at the level of resolution of our sensory acuity (the "jnd" or just-noticeable-difference), then there is always a region of uncertainty at the border of a category, dependent on the accuracy and sensitivity of the threshold-detector. But discrimination grain is not the right level of analysis for questions about higher-order sensory categories, and all-or-none categorization in general. The case for the putative "gradedness" of "penguin"'s membership in the category "bird" is surely not being based on the limits of sensory acuity. If it is, I'll concede at once, and add that that sort of gradedness is trivial; the categorization problem is concerned with identification grain, not discrimination grain. All categories will of course be fuzzy at the limits of our sensory resolution capacity. My own grounding hypothesis BEGINS with bounded sensory categories (modulo threshold uncertainty) and attempts to ground the rest of our category hierarchy bottom-up on those. Finally, as I've stressed in responses to others, there's one other form of category uncertainty I'm quite prepared to concede, but that likewise fails to imply that category membership is a matter of degree: All categories -- true graded ones as well as all-or-none ones -- are provisional and approximate, relative to the context of interconfusable members and nonmembers that have been sampled to date. If the sample ever turns out to have been nonrepresentative, the feature-set that was sufficient to generate successful sorting in the old context must be revised and updated to handle the new, wider context. Anomalies and ambiguities that had never occurred before must now be handled. But what happens next (if all-or-none sorting performance can be successfully re-attained at all) is just the same as with the initial category learning in the old context: A set of features must be found that is sufficient to subserve correct performance in the extended context. The approximation must be tightened. This open-endedness of all of our categories, however, is really just a symptom of inductive risk rather than of graded representations. > "Analog" means "invertible". The invertible properties of a > representation are those properties which it preserves...[This > sounds] tautologically true of *all* representations. For the reply to this, see my response to Cugini, whose criticism you cite. Sensory icons need only be invertible with the discriminable properties of the sensory projection. There is no circularity in this. And in a dedicated system invertibility at various stages may well be a matter of degree, but this has nothing to do with the issue of graded/nongraded category membership, which is much more concerned with selective NONinvertibility. > It is quite possible to make all-or-none judgements based on graded > features [e.g., thermostats] Apart from (1) thresholds (which are features, and which I discussed earlier), (2) probabilistic features so robust as to be effectively all-or-none, and (3) gerrymandered examples (usually playing on the finiteness of the cases sampled, and the underdetermination of the winning feature set), can you give examples? > "chair"... with no back... [is a] "stool"... Now vary the size > of the back The linguist Labov, with examples such as cup/bowl, specialized in finding graded regions for seemingly all-or-none categories. Categorization is always a context-dependent, "compared-to-what" task . Features must reliably sort the members from the nonmembers they can be confused with. Sometimes nature cooperates and gives us natural discontinuities (horses could have graded continuously into zebras). Where she does not, we have only one recourse left: an all-or-none sensory threshold at some point in the continuum. One can always generate a real or hypothetical continuum that would foil our current feature-detectors and necessitate a threshold-detector. Such cases are only interesting if they are representative of the actual context of confusable alternatives that our category representation must resolve. Otherwise they are not informative about our actual current (provisional) feature-set. > I see nothing about "all-or-none" categories which is not explainable > by arbitrary cutoffs of graded sensory data... [and] avoid[ing] > borderline situations. Neither do I. (Most feature-detection problems, by the way, do not arise from the need to place thresholds along true continua, but from the problem of underdetermination: there are so many features that it is hard to find a set that will reliably sort the confusable alternatives into their proper all-or-none categories.) -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 5 Jul 87 00:51:01 GMT From: sher@cs.rochester.edu (David Sher) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <967@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: > >Most of our object categories are indeed all-or-none, not graded. A >penguin is not a bird as a matter of degree. It's a bird, period. Just for the record is this an off hand statement or are you speaking as an expert when you say most of our categories are all or none. Do you have some psychology experiments that measure the size of human category spaces and using a metric on them shows that most categories are of this form? Can I quote you on this? Personally I have trouble imagining how to test such a claim but psychologists are clever fellows. -- -David Sher sher@rochester { seismo , allegra }!rochester!sher ------------------------------ Date: 5 Jul 87 04:52:30 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In Article 185 of comp.cog-eng sher@rochester.arpa (David Sher) of U of Rochester, CS Dept, Rochester, NY responded as follows to my claim that "Most of our object categories are indeed all-or-none, not graded. A penguin is not a bird as a matter of degree. It's a bird, period." -- > Personally I have trouble imagining how to test such a claim... Try sampling concrete nouns in a dictionary. -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 5 Jul 87 05:29:02 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem In Article 184 of comp.cog-eng: adam@gec-mi-at.co.uk (Adam Quantrill) of Marconi Instruments Ltd., St. Albans, UK writes: > It seems to me that the Symbol Grounding problem is a red herring. > If I took a partially self-learning program and data (P & D) that had > learnt from a computer with 'sense organs', and ran it on a computer > without, would the program's output become symbolically ungrounded?... > [or] if I myself wrote P & D without running it on a computer at all? This begs two of the central questions that have been raised in this discussion: (1) Can one speak of grounding in a toy device (i.e., a device with performance capacities less than those needed to pass the Total Turing Test)? (2) Could the TTT be passed by just a symbol manipulating module connected to transducers and effectors? If a device that could pass the TTT were cut off from its transducers, it would be like the philosophers' "brain in a vat" -- which is not obviously a digital computer running programs. -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 5 Jul 87 02:47:25 GMT From: ihnp4!twitch!homxb!houdi!marty1@ucbvax.Berkeley.EDU (M.BRILLIANT) Subject: Re: The symbol grounding problem In article <605@gec-mi-at.co.uk>, adam@gec-mi-at.co.uk (Adam Quantrill) writes: > It seems to me that the Symbol Grounding problem is a red herring. As one who was drawn into a problem that is not my own, let me try answering that disinterestedly. To begin with, a "red herring" is something drawn across the trail that distracts the pursuer from the real goal. Would Adam tell us what his real goal is? Actually, my own real goal, from which I was distracted by the symbol grounding problem, was an expert system that would (like Adam's last example) ground its symbols only in terminal I/O. But that's a red herring in the symbol grounding problem. ..... If I took a partially self-learning program and data (P & D) that had learnt from a computer with 'sense organs', and ran it on a computer without, would the program's output become symbolically ungrounded? No, because the symbolic data was (were?) learned from sensory data to begin with - like a sighted person who became blind. Similarily, if I myself wrote P & D without running it on a computer at all, [and came] up with identical P & D by analysis. Does that make the original P & D running on the computer with 'sense organs' symbolically ungrounded? No, as long as the original program learned its symbolic data from its own sensory data, not by having them defined by a person in terms of his or her sensory data. A computer can always interact via the keyboard & terminal screen, (if those are the only 'sense organs'), grounding its internal symbols via people who react to the output, and provide further stimulus. That's less challenging and less useful than true symbol grounding. One problem that requires symbol grounding (more useful and less ambitious than the Total Turing Test) is a seeing-eye robot: a machine with artificial vision that could guide a blind person by giving and taking verbal instructions. It might use a Braille keyboard instead of speech, but the "terminal I/O" must be "grounded" in visual data from, and constructive interaction with, the tangible world. The robot could learn words for its visual data by talking to people who could see, but it would still have to relate the verbal symbols to visual data, and give meaning to the symbols in terms of its ultimate goal (keeping the blind person out of trouble). M. B. Brilliant Marty AT&T-BL HO 3D-520 (201)-949-1858 Holmdel, NJ 07733 ihnp4!houdi!marty1 ------------------------------ End of AIList Digest ******************** 6-Jul-87 01:05:39-PDT,14968;000000000000 Mail-From: LAWS created at 6-Jul-87 01:04:24 Date: Mon 6 Jul 1987 01:02-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #170 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 6 Jul 1987 Volume 5 : Issue 170 Today's Topics: Theory - Symbol Grounding Metadiscussion ---------------------------------------------------------------------- Date: 3 Jul 87 01:02:48 GMT From: mnetor!utzoo!utgpu!water!watmath!watcgl!ksbooth@seismo.css.gov Subject: Re: The symbol grounding problem - please start your own newsgroup Hooray for David Harwood. ------------------------------ Date: 5 Jul 87 05:39:38 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem - please start your own newsgroup In Article 186 of comp.cog-eng, ksbooth@watcgl.waterloo.edu (Kelly Booth) of U. of Waterloo, Ontario writes: > Hooray for David Harwood. David Harwood has made two very rude requests that I stop the symbol grounding discussion, which I ignored. But perhaps it's time to take a poll. Please send me e-mail indicating whether or not you find the discussion useful and worth continuing. I promise to post and abide by the results. -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 5 Jul 87 05:05:53 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: The symbol grounding problem: 3 routes to grounding needed? In Article 181 of comp.cog-eng berleant@ut-sally.UUCP (Dan Berleant) of U. Texas CS Dept., Austin, Texas writes: > may not be much difference between a classical view augmented to... > *arbitrary* boolean expressions of features...and a probabilistic view I agree that such a probabilistic representation is possible. Now the question is, will it work, is it economical (and is it right)? Note, though, that even graded (probabilistic) individual features must yield an all-or-none feature SET. So even this would not be evidence of graded membership. (I don't think you'd disagree.) > need to...explain...typicality and reaction time results...interpreted > as supporting probabilistic and exemplar-based category representations Yes, but it seems only appropriate that we should account for the categorization performance capacity itself before we worry about its fine tuning. (Experimental psychology has a long history of bypassing the difficult but real problems underlying our behavioral capacities and fixating instead on fine-tuning.) > may [be] 2 representations for categories: a 'core' of defining features > and a heuristic categorizer... 2 pathways [grounding] categories You may be right. It's an empirical question whether the heuristic component will be necessary to generate successful performance. If it is, it is still not obvious that the need for it would be directly related to the grounding problem. > [Re:] Anders Weinstein [on] the semantic meaning of...thunder/...`angry > gods nearby'...: The terms in the definition presumably are grounded > via the 2 routes discussed above... [now] Consider a sentence with 2 > variables, e.g. FISH SWIM... Obviously, many bindings would satisfy > the sentence. [But]...by adding many more true sentences, the possible > bindings of the variables become much more constrained. I accepted this argument the first time you made it. I think it's right; I've made similar degrees-of-freedom arguments against Quine myself, and I've cross-referenced your point in my response to Weinstein. I don't believe, though, that this reduction of the degrees of freedom of the interpretation (even to zero) is sufficient to ground a symbol system. Even if there's only one way to interpret an entire language, the decryption must be performed; and it's not enough that the mapping should be into a natural language (that's still a symbol/symbol relation, leaving the entire edifice hanging by a skyhook of derived rather than intrinsic meaning). The mapping must be into the world. But, in any case, you seem to rescind your degrees-of-freedom argument immediately after you make it: > On the other hand... Maybe a Martian [or] your neighbor... could > figure out [an alternative] way to do it consistently... but as long > as you both agree on the truthfulness of all the sentences you are > mutually aware of, there is no way to tell! Shades of the Turing test... This is standard Quinean indeterminacy again! So you don't believe your degrees-of-freedom argument! Well I do. And it's partly because of degrees-of-freedom and convergence considerations that I am so sanguine about the TTT. (I called this the "convergence" argument in "Minds, Machines and Searle": There may be many arbitrary ways to successfully model a toy performance, but as you move toward the TTT, the degrees of freedom shrink.) > would this method of 'grounding' the semantics of categories be > sufficient to do the job? Only in theory? Potentially in practice? ... I think it would not (although it may simplify the task of grounding somewhat). Even if only one interpretation is possible, it must be intrinsic, not derivative. > Are you assuming a representation of episodes (more generally, > exemplars) that is iconic rather than symbolic? Yes, I am assuming that episodic representations would be iconic. This is related to the distinction in the human memory literature concering "episodic" vs. "semantic" memory. The former involves qualitative recall for when something happened (e.g., Kennedy's assassination) and the particulars of the experience; the latter involves only the *product* of past learning (e.g., knowing how to ride a bicycle, do calculus or speak English). It's much harder to imagine how the former could be symbolic (although, of course, there are "constructive" memory theories such as Bartlett's that suggest that what we remember as an episode may be based on reconstruction and logical inference...). > *no* category representation method can generate category boundaries > when there is significant interconfusability among categories! I would be very interested to know your basis for this assertion (particularly as "significant interconfusability" is not exactly a quantitative predicate). If I had said "complete indeterminacy," or even "radical underdetermination" (say, features that would require exponential search to find), I could understand why you would say this -- but significant interconfusability... Can you remember first looking at cellular structures under a microscope? Have you seen Inuit snow taxonomies? Have you ever tried serious mushroom-picking? Or chicken sexing? Or tumor identification? Art classification? Or, to pick some more abstract examples: paleolinguistic taxonomy? ideological typologizing? or problems at the creative frontiers of pure mathematics? -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 5 Jul 87 18:34:37 GMT From: bloom-beacon!bolasov@husc6.harvard.edu (Benjamin I Olasov) Subject: Re: The symbol grounding problem - please start your own newsgroup I personally don't feel that it's Harwood's place to make a recommendation such as the one he made (rude or otherwise). If the discussion is germaine to the stated purpose(s) of the newsgroup (which it is), and is carried on in an intellectually responsible manner (which it certainly has been), why should it not be allowed to continue? Isn't the solution for those who don't find the topic interesting to simply not read the messages bearing that topic on the subject line? After all, any number of discussions can be carried on concurrently. ------------------------------ Date: 5 Jul 87 17:31:15 GMT From: harwood@cvl.umd.edu (David Harwood) Subject: Re: The symbol grounding problem - please start your own newsgroup In article <977@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: > >David Harwood has made two very rude requests that I stop the symbol grounding >discussion, which I ignored. But perhaps it's time to take a poll. Please send >me e-mail indicating whether or not you find the discussion useful and worth >continuing. I promise to post and abide by the results. As I have told others, I don't really want you to quit posting altogether to this or other newsgroups. And I would be glad for you to form your own group for your "dialogues," such as they are. But I have to complain about your insufferable postings on two grounds: (i) they have nearly nothing to do with computer science, nevertheless preoccupy comp.ai with your various and sundry self-referential, just vaguely intelligible musings; (ii) your postings, in my opinion, are the heighth, width, and breadth of unresponsive, presumptuous, and condescending twaddle. Worse than anything which I've read which was contributed as an original article to BBS, for example. Of course, as my colleagues advise, BBS does not publish my research - and is unlikely to in the near distant future. Such are the wages of public sin.) Yes, my two replies to you were sarcastic (more than "very rude," I think; I never recieved any serious complaint about either, perhaps because others knew what I meant, even if they did not quite agree with me.) Let me give you back an illustration of how you talk. You just a moment ago replied to D.S. who question what psychological evidence you have that perceptual categorization is usually "all-or-none." He seemed to question your expertese as a perceptual psychologist. (I might add that you have tried to impress us with generally slighting remarks about psychologists as well as computer scientists, but this may be a "policy of controversy" (perhaps used to secure competitive funding - who knows;-). Anyway, your one line reply did not answer the question, but was more of a silly riposte, something like, "Check the concrete nouns in your dictionary." He asks you something, and you ignore this. Or, taking you seriously, you tell him to go supply his own evidence for your claims. (I suppose that if he were your research assistant, that you would sagely explain that a "concrete" noun is one admitting "all-or-none" categorization.) I have no prejudice concerning your views - to be sure, I rarely can make sense of them. But I wish you would simply take your own advise, "Check the concrete nouns of your dictionary," and use them sometimes to good effect in your postings. Define your abstractions. Cite evidence for your speculations. Do not cite your own damn article like a parrot. If you prefer, post the damn thing, which has got to be more intelligible than your recent stuff, and we will be done with this particular "symbol grounding problem." Then I will look forward to your new occasional postings, even in this newsgroup. David Harwood ------------------------------ Date: 5 Jul 87 21:48:28 GMT From: harwood@cvl.umd.edu (David Harwood) Subject: Re: The symbol grounding problem - please start your own newsgroup Letter sent by email to Stevan Harnad (with postscript added) re his postings to comp.ai about "the symbol grounding problem." I don't want you to quit posting altogether - I would just like you to realize that you are hogging comp.ai with what seems, to me at least, to be mostly pompous and unintelligible postings, that have very little to do with computer science. I heard from a student colleague, who is not opposed to a "cognitive science" viewpoint (if this means anything to you), that the first thing you did to explain your views at a recent colloquim was make reference to your net discussions. My oh my, either you are an modest comedian, or these dialogues of yours - why - if even they be blarney and posing of feathers - why they be verily verily immortal. You have made your views, whatever these are, resoundingly reknown - by, I suppose, half or more of the recent volume of comp.ai. I simply wish you'd pipe down for awhile, especially about your "symbol grounding problem." I will be especially verily verily glad to see you post the source code which implements your theoretical improvements; this should keep us off the streets for awhile; and I will try to be first to applaud your success. David Harwood Computer Vision Laboratory Center for Automation Research University of Maryland My views are simply my own. Please note all typos and mistakes, as I prepare to publish an edition (with permission which is surely forthcoming) of _Recent Contributions to the Dialogue de Problem Profundo Symbo-Grundo: New Foundations and New Vocations in Computer Science_. [This postscript added to my letter emailed S.H.] ------------------------------ Date: 6 Jul 87 02:19:01 GMT From: bloom-beacon!bolasov@husc6.harvard.edu (Benjamin I Olasov) Subject: Re: The symbol grounding problem - please start your own newsgroup In article <2328@cvl.umd.edu> harwood@cvl.UUCP (David Harwood) writes: > I don't want you to quit posting altogether - I would just like >you to realize that you are hogging comp.ai with what seems, to me at least, >to be mostly pompous and unintelligible postings, that have very little to >do with computer science. ^^^^^^^^ ^^^^^^^ This point should not need to be made, but this newsgroup doesn't deal exclusively with computer science issues per se. Many important contributions to AI, after all, have come from outside the field of CS, as conventionally understood- much of Marvin Minsky's research for example, is not restricted to CS, and yet has significant implications for AI. Some of the most challenging and interesting problems of AI are philosophical in nature. I frankly don't see why this fact should disturb anyone. Perhaps if more of us pursued our theoretical models with comparable rigor to that with which Mr. Harnad pursues his, the balance of topics represented on comp.ai might shift ..... ------------------------------ End of AIList Digest ******************** 6-Jul-87 01:11:40-PDT,14518;000000000001 Mail-From: LAWS created at 6-Jul-87 01:09:35 Date: Mon 6 Jul 1987 01:07-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #171 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 6 Jul 1987 Volume 5 : Issue 171 Today's Topics: Queries - AI Expert Source for Hopfield Nets & Liability of Expert System Developers, Programming - Software Reuse, Scientific Method - Psychology vs. AI & Why AI is not a Science ---------------------------------------------------------------------- Date: 2 Jul 87 20:45:24 GMT From: ucsdhub!dcdwest!benson@sdcsvax.ucsd.edu (Peter Benson) Subject: AI Expert source for Hopfield Nets I am looking for the source mentioned in Bill Thompson's article on Hopfield Nets in the July, 1987 issue of AI Expert magazine. At one time, someone was posting all the sources, but has, apparently, stopped. Could that person, or some like-minded citizen post the source for this Travelling Salesman solution. Thanks in advance !! -- Peter Benson | ITT Defense Communications Division (619)578-3080 | 10060 Carroll Canyon Road ucbvax!sdcsvax!dcdwest!benson | San Diego, CA 92131 dcdwest!benson@SDCSVAX.EDU | ------------------------------ Date: 5 Jul 87 22:00:58 GMT From: bloom-beacon!bolasov@husc6.harvard.edu (Benjamin I Olasov) Subject: Liability of Expert System Developers I'm told that a hearing is now underway which would set a legal precedent for determining the extent of liability to be borne by software developers for the performance of expert systems authored by them. Does anyone have details on this? ------------------------------ Date: 4 Jul 87 21:19:48 GMT From: jbn@glacier.stanford.edu (John B. Nagle) Subject: Re: Software Reuse -- do we really know what it is ? (long) The trouble with this idea is that we have no good way to express algorithms "abstractly". Much effort was put into attempting to do so in the late 1970s, when it looked as if program verification was going to work. We know now that algebraic specifications (of the Parnas/SRI type) are only marginally shorter than the programs they specify, and much less readable. Mechanical verification that programs match formal specifications turned out not to be particularly useful for this reason. (It is, however, quite possible; a few working systems have been constructed, including one by myself and several others described in ACM POPL 83). We will have an acceptable notation for algorithms when each algorithm in Knuth's "Art of Computer Programming" is available in machineable form and can be used without manual modification for most applications for which the algorithm is applicable. As an exercise for the reader, try writing a few of Knuth's algorithms as Ada generics and make them available to others, and find out if they can use them without modifying the source text of the generics. In practice, there now is a modest industry in reusable software components; see the ads in any issue of Computer Language. Worth noting is that most of these components are in C. John Nagle ------------------------------ Date: 02 Jul 87 09:55:35 EDT (Thu) From: sas@bfly-vax.bbn.com Subject: Don Norman's comments on time perception and AI philosophizing Actually, many studies have been done on time perception. One rather interesting one reported some years back in Science showed that time and size scale together. Smaller models (mannikins in a model office setting) move faster. It was kind of neat paper to read. I agree that AI suffers from a decidedly non-scientific approach. Even when theoretical physicists flame about liberated quarks and the anthropic principle, they usually have some experiments in mind. In the AI world we get thousands of bytes on the "symbol grounding problem" and very little evidence that symbols have anything to do with intelligence and thought. (How's that for Drano[tm] on troubled waters?) There have been a lot of neat papers on animal (and human) learning coming out lately. Maybe the biological brain hackers will get us somewhere - at least they look for evidence. Probably overstating my case, Seth ------------------------------ Date: Thu 2 Jul 87 12:10:08-PDT From: PAT Subject: Re: AIList Digest V5 #165 HEY, DON!!! RIGHT ON! Pat Hayes [Donald Norman, I presume. -- KIL] ------------------------------ Date: 3 Jul 87 18:01:33 GMT From: nosc!humu!uhccux!stampe@sdcsvax.ucsd.edu (David Stampe) Subject: Submission for comp-ai-digest Path: uhccux!stampe From: stampe@uhccux.UUCP (David Stampe) Newsgroups: comp.ai.digest Subject: Re: On how AI answers psychological issues Message-ID: <651@uhccux.UUCP> Date: 3 Jul 87 18:01:33 GMT References: <8706301418.AA08078@sunl.ICS> Distribution: world Organization: U. of Hawaii, Manoa (Honolulu) Lines: 44 In-reply-to: norman%ics@SDCSVAX.UCSD.EDU's message of 30 Jun 87 14:18:40 GMT norman%ics@SDCSVAX.UCSD.EDU (Donald A. Norman) writes: > Thinking about "how the mind works" is fun, but not science, not > the way to get to the correct answer. In fact it's the ONLY way to get the correct answer. Experiments don't design themselves, and they don't interpret their own results. We don't see with outward eyes or hear with outward ears alone. The outward perception or behavior does not exist without the inward one. If you practice your remembered violin in your imagination, while your actual violin is being repaired, you, as well as the violin, may sound much better when the repairs are finished. I am a linguist. I write a tongue twister on the board that they haven't hear before: 'Unique New York Unique New York Unique New York....' My students watch silently, but when I ask them what errors this tongue twister induces, they immediately name the very errors I discovered before class, when I tried to pronounce it aloud. You didn't have to say it aloud, either, did you? It is not introspection that is AI's trouble. It is that an expert system, for example, isn't likely to model expertise correctly until it is designed by someone who is himself the expert, or who knows how to discover the nature of the expert's typically unconscious wisdom. Linguistics has struggled for over a century to develop tools for learning how human beings acquire and use language. It seems likely that a comparable struggle will be required learn how the expert diagnostician, welder, draftsman, or reference librarian does what he or she does. I often feel that when a good student of language takes a job building a natural language interface for some AI project, in her work -- though it may be viewed by others in the project as marginal, if not menial -- she is more likely to turn up something of scientific import than are those working on the core of the project. This is just because she has spent years learning to learn how experts -- in this case language users -- do what they do. On the other hand, she is not likely to believe that programs can realistically model much of the human linguistic faculty, at least in the imaginable future. For example, computer parsers presuppose grammars. But it is not clear whether children, the only devices so far known to have mastered any natural language, come equipped with any analogous utilities. David Stampe, Linguistics, Univ. of Hawaii ------------------------------ Date: Thu, 2 Jul 87 22:36:05 edt From: amsler@flash.bellcore.com (Robert Amsler) Subject: Re: thinking about thinking not being science I think Don Norman's argument is true for cognitive psychologists, but may not be true for AI researchers. The reason is that the two groups seek different answers. If AI were only the task of finding out how people work, then it would be valid to regard armschair reasoning as an invalid form of speculation. One can study people directly (this is the old ``stop arguing over the number of teeth in a horse's mouth and go outside and count them'' argument). However, some AI researchers are really engineers at heart. The question then is not how do people work, but how could processes providing comparable performance quality to those of humans be made to work in technological implementations. `Could' is important. Airplanes are clearly not very good imitations of birds. They are too big, for one thing. They have wheels instead of feet, and the list goes on and on (no feathers!). Speculating about flight might lead to building other types of aircraft (as certainly those now humorous old films of early aviation experiments show), but it would certainly be a bad procedure to follow to understand birds and how they fly. Speculating about why the $6M man appears as he does while running is a tad off the beaten path for AILIST, but that process of speculation is hardly worthless for arriving at novel means of representing memory or perception FOR COMPUTER SYSTEMS. Lets not squabble over the wrong issue. The problem is that the imagery of the $6M man's running is just too weak as a springboard for much directed thought and the messages (including my own earlier reply) are just rambling off in directions more appropriate to SF-Lovers than AILIST. I do agree that the CURRENT discussion isn't likely to lead anywhere--but not that the method of armchair speculation is invalid in AI. ------------------------------ Date: Fri, 3 Jul 87 07:29:41 pdt From: norman%ics@sdcsvax.ucsd.edu (Donald A. Norman) Subject: Why AI is not a science A private message to me in response to my recent AI List posting, coupled with general observations lead me to realize why so many of us otherwise friendly folks in the sciences that neighbor AI can be so frustrated with AI's casual attitude toward theory: AI is not a science and its practitioners are woefuly untutored in scientific method. At the recent MIT conference on Foundations of AI, Nils Nilsson stated that AI was not a science, that it had no empirical content, nor claims to emperical content, that it said nothing of any emperical value. AI, stated Nilsson, was engineering. No more, no less. (And with that statement he left to catch an airplane, stopping further discussion.) I objected to the statement, but now that I consider it more deeply, I believe it to be correct and to reflect the dissatisfaction people like me (i.e., "real scientists") feel with AI. The problem is that most folks in AI think they are scientists and think they have the competence to pronounce scientific theories about almost any topic, but especially about psychology, neuroscience, or language. Note that perfectly sensible dsciplines such as mathematics and philosophy are also not sciences, at least not in the normal intrerpretation of that word. It is no crime not to be a science. The crime is to think you are one when you aren't. AI worries a lot about methods and techniques, with many books and articles devoted to these issues. But by methods and techniques I mean such topics as the representation of knowledge, logic, programming, control structures, etc. None of this method includes anything about content. And there is the flaw: nobody in the field of Artificial Intelligence speaks of what it means to study intelligence, of what scientific methods are appropriate, what emprical methods are relevant, what theories mean, and how they are to be tested. All the other sciences worry a lot about these issues, about methodology, about the meaning of theory and what the appropriate data collection methods might be. AI is not a science in this sense of the word. Read any standard text on AI: Nilsson or Winston or Rich or even the multi-volumned handbook. Nothing on what it means to test a theory, to compare it with others, nothing on what constitutes evidence, or with how to conduct experiments. Look at any science and you will find lots of books on experimental method, on the evaluation of theory. That is why statistics are so important in psychology or biology or physics, or why counterexamples are so important in linguistics. Not a word on these issues in AI. The result is that practitioners of AI have no experience in the complexity of experimental data, no understanding of scientific method. They feel content to argue their points through rhetoric, example, and the demonstration of programs that mimic behavior thought to be relevant. Formal proof methods are used to describe the formal power of systems, but this rigor in the mathematical analysis is not matched by any similar rigor of theoretical analysis and evaluation for the content. This is why other sciences think that folks in AI are off-the-wall, uneducated in scientific methodology (the truth is that they are), and completely incompetent at the doing of science, no matter how brilliant at the development of mathematics of representation or formal programming methods. AI will contribute to the A, but will not contribute to the I unless and until it becomes a science and develops an appreciation for the experimental methods of science. AI might very well develop its own methods -- I am not trying to argue that existing methods of existing sciences are necessarily appropriate -- but at the moment, there is only clever argumentation and proof through made-up example (the technical expression for this is "thought experiment" or "gadanken experiment"). Gedanken experiments are not accepted methods in science: they are simply suggestive for a source of ideas, not evidence at the end. don norman Donald A. Norman Institute for Cognitive Science C-015 University of California, San Diego La Jolla, California 92093 norman@nprdc.arpa {decvax,ucbvax,ihnp4}!sdcsvax!ics!norman norman@sdics.ucsd.edu norman%sdics.ucsd.edu@RELAY.CS.NET ------------------------------ End of AIList Digest ******************** 9-Jul-87 19:40:41-PDT,19966;000000000001 Mail-From: LAWS created at 8-Jul-87 17:29:53 Date: Wed 8 Jul 1987 17:22-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #172 To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 9 Jul 1987 Volume 5 : Issue 172 Today's Topics: Query - Xlisp, Programming - Software Reuse & Abstract Specifications, Scientific Method - Is AI a Science? ---------------------------------------------------------------------- Date: Tue 7 Jul 87 09:18:25-PDT From: BEASLEY@EDWARDS-2060.ARPA Subject: xlisp If anyone has any information or has heard any information about using XLISP (eXperimental LISP) on the PC, please send me that information at beasley@edwards-2060.ARPA. Thank you. ------------------------------ Date: 6 Jul 87 05:28:23 GMT From: vrdxhq!verdix!ogcvax!dinucci@seismo.css.gov (David C. DiNucci) Subject: Re: Software Reuse -- do we really know what it is ? (long) In article ijd@camcon.co.uk (Ian Dickinson) writes: >> Xref: camcon comp.lang.ada:166 comp.lang.misc:164 >Hence a solution: we somehow encode _abstractions_ of the ideas and place >these in the library - in a form which also supplies some knowledge about the >way that they should be used. The corollary of this is that we need more >sophisticated methods for using the specifications in the library. >(Semi)-automated transformations seem to be the answer to me. > >Thus we start out with a correct (or so assumed) specification, apply >correctness-preserving transormation operators, and so end up with a correct >implementation in our native tongue (Ada, Prolog etc, as you will). The >transformations can be interactively guided to fit the precise circumstance. >[Credit] I originally got this idea from my supervisor: Dr Colin Runciman >@ University of York. In his Phd thesis defense here at Oregon Graduate Center, Dennis Volpano presented his package that did basically this. Though certainly not of production quality, the system was able to take an abstraction of a stack and, as a separate module, a description of a language and data types within the language (in this case integer array and file, if I remember correctly), and produce code which was an instantiation of the abstraction - a stack implemented as an array or as a file. I haven't actually read Dennis' thesis, so I don't know what the limitations of constraints on his approach are. I believe he is currently employed in Texas at MCC. --- Dave DiNucci dinucci@Oregon-Grad ------------------------------ Date: 7 Jul 87 02:21:06 GMT From: vrdxhq!verdix!ogcvax!pase@seismo.css.gov (Douglas M. Pase) Subject: Re: Software Reuse (short title) In article jbn@glacier.UUCP (John B. Nagle) writes: > > The trouble with this idea is that we have no good way to express >algorithms "abstractly". [...] Well, I'm not sure just where the limits are, but polymorphic types can go a long way towards what you have been describing. It seems that a uniform notation for operators + the ability to define additional operators + polymorphically typed structures are about all you need. Several functional languages already provide an adequate basis for these features. One such language is called LML, or Lazy ML. Current language definitions tend to concentrate on the novel features rather than attempt to make LML a full-blown "production" language, and therefore may be missing some of your favorite features. However, my point is that we may well be closer to your objective than some of us realize. I apologize for the brevity of this article -- if I have been too vague, send me e-mail and I will be more specific. -- Doug Pase -- ...ucbvax!tektronix!ogcvax!pase or pase@Oregon-Grad.csnet ------------------------------ Date: 7 Jul 87 15:18:32 GMT From: debray@arizona.edu (Saumya Debray) Subject: Automatic implementation of abstract specifications In article <1337@ogcvax.UUCP>, dinucci@ogcvax.UUCP (David C. DiNucci) writes: > In his Phd thesis defense here at Oregon Graduate Center, Dennis > Volpano presented his package that did basically this. Though certainly > not of production quality, the system was able to take an abstraction > of a stack and, as a separate module, a description of a language and > data types within the language (in this case integer array and file, > if I remember correctly), and produce code which was an instantiation > of the abstraction - a stack implemented as an array or as a file. I believe there was quite a bit of work on this sort of stuff at MIT earlier in the decade. E.g. there was a PhD thesis [ca. 1983] by M. K. Srivas titled "Automatic Implementation of Abstract Data Types" (or something close to it). The idea, if I remember correctly, was to take sets of equations specifying the "source" ADT (e.g. stack) and the "target" ADT (e.g. array), and map the source into the target. -- Saumya Debray CS Department, University of Arizona, Tucson internet: debray@arizona.edu uucp: {allegra, cmcl2, ihnp4} !arizona!debray ------------------------------ Date: Mon, 6 Jul 87 10:06:05 MDT From: shebs%orion@cs.utah.edu (Stanley T. Shebs) Subject: AI vs Scientific Method I can understand Don Norman's unhappiness about the lack of scientific method in AI - from a practical point of view, the lack of well-understood criteria for validity means that refereeing of publications is unlikely to be very objective... :-( The scientific method is a two-edged sword, however. Not only does it define what is interesting, but what is uninteresting - if you can't devise a con- trolled experiment varying just a single parameter, you can't say anything about a phenomenon. A good scientist will perhaps be able to come up with a different experiment, but if stymied enough times, he/she is likely to move on to something else (at about the same time the grant money runs out :-) ). Established sciences like chemistry have an advantage in that the parameters most likely to be of interest are already known; for instance temperature, percentages of compounds, types of catalysts, and so forth. What do we have for studying intelligence? Hardly anything! Yes, I know psychologists have plenty of experimental techniques, but the quality is pretty low compared to the "hard sciences". A truly accurate psychology experiment would involve raising cloned children in a computer-controlled environment for 18 years. Even then, you're getting minute amounts of data about incredibly complex systems, with no way to know if the parameters you're varying are even relevant. There's some consolation to be gained from the history of science/technology. The established fields did not spring full-blown from some genius' head; each started out as a confused mix of engineering, science, and speculation. Most stayed that way until the late 19th or early 20th century. If you don't believe me, look at an 18th or early 19th century scientific journal (most libraries have a few). Quite amusing, in fact very similar to contemporary AI work. For instance, an article on electric eels from about 1780 featured the observations that a slave grabbing the eel got a stronger shock on the second grab, and that the shock could be felt through a wooden container. No tables or charts or voltmeter readings :-). My suggestion is to not get too worked up about scientific methods in AI. It's worth thinking about, but people in other fields have spent centuries establishing their methods, and there's no reason to suppose it will take any less for AI. stan shebs shebs@cs.utah.edu ------------------------------ Date: Mon, 6 Jul 1987 16:29 EDT From: MINSKY%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: AIList Digest V5 #170 I would like to see that discussion of "symbol grounding" reduced to much smaller proportions because I think it is not very relevant to AI, CS, or psychology. To understand my reason, you'd have to read "Society of Mind", which argues that this approach is obsolete because it recapitulates the "single agent" concept of mind that dominates traditional philosophy. For example, the idea of "categorizing" perceptions is, I think, mainly an artifact of language; different parts of the brain deal with inputs in different ways, in parallel. In SOM I suggest many alternative ways to think about thinking and, in several sections, I also suggest reasons why the single agent idea has such a powerful grip on us. I realize that it might seem self-serving for me to advocate discussing Society of Mind instead. I would have presented my arguments in reply to Harnad, but they would have been too long-winded and the book is readily available. ------------------------------ Date: Mon, 6 Jul 87 18:25:51 EDT From: Jim Hendler Subject: Re: AIList Digest V5 #171 While I have some quibbles with Don N.'s long statement on AI viz (or vs.) science, I think he gets close to what I have felt a key point for a long time -- that the move towards formalism in AI, while important in the change of AI from a pre-science (alchemy was Drew McDermott's term) to a science, is not enough. For a field to make the transition an experimental methodology is needed. In AI we have the potential to decide what counts as experimentation (with implementation being an important consideration) but have not really made any serious strides in that direction. When I publish work on planning and claim ``my system makes better choices than '' I cannot verify this other than by showing some examples that my system handles that 's can't. But of course, there is no way of establishing that couldn't do examples mine can't and etc. Instead we can end up forming camps of beliefs (the standard proof methodology in AI) and arguing -- sometimes for the better, sometimes for the worse. While I have no solution for this, I think it is an important issue for consideration, and I thank Don for provoking this discussion. -Jim Hendler ------------------------------ Date: Tue, 7 Jul 1987 01:11 EDT From: MINSKY%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: AIList Digest V5 #171 At the end of that long and angry flame, I think D.Norman unwittingly hit upon what made him so mad: > Gedanken experiments are not accepted methods in science: they are > simply suggestive for a source of ideas, not evidence at the end. And that's just what AI has provided these last thirty years - a source of ideas that were missing from psychology in the century before. Representation theories, planning procedures, heuristic methods, hundreds of such. The history of previous psychology is ripe with "proved" hypotheses, few of which were worth a damn, and many of which were refuted by Norman himself. Now "cognitive psychology" - which I claim and Norman will predictably deny (see there: a testable hypothesis!) is largely based on AI theories and experiments - is taking over at last - as a result of those suggestions for ideas. ------------------------------ Date: Tue, 7 Jul 87 01:28 MST From: "Paul B. Rauschelbach" Subject: What is science I normally only observe this discussion, but Don Norman's pomposity struck a nerve. The first objection I have is to his statement that mathematics and philosophy are not sciences "in the normal interpretation of the word." The Webster's definition (a fairly normal interpretation) is: "accumulated knowledge systematized and formulated with reference to the discovery of general truths or the operation of general laws." This certainly applies to both. The next problem is his statement that AI people think they're scientists. He seemed to believe that it was a science until Nils Nilsson told him the obvious. AI, like it's name implies, is a product, not a phenomenon, not an occurence of nature to be described. The problem is the creation of a product, an engineering problem. The preservation of theory is far from an engineer's mind. The engineer uses theory to describe possible solutions. If an engineer comes across a possible solution that has not been addressed by theory, s/he may get his hands a little dirty before the "scientists" take control of it. It seems to me that much of the talk in this discussion is of a hypothetical nature, one of the elements of THE SCIENTIFIC METHOD he was defending. This is a good place for that portion of the method, as well as statement of the problem. The experimentation is left to the psychologists, neurologists, etc. I see no one but scientists claiming to be scientists, and I hear AI people shouting, "Yeah, but how do you code it?" or "What doohickey will do that?" Implementation of theory. I have also read discussion of the testing of implementation. Come to think of it, engineering also fits the definition of science. Both things, implementation and theory have been and should be discussed here. If they intermingle, this can only be healthy, even if somewhat confusing. I hope we can both get down off our respective high horses now. Paul Rauschelbach Honeywell Bull P.O. Box 8000, M/S K55, Phoenix, AZ 85006 (602) 862-3650 pbr%pco@BCO-MULTICS.ARPA Disclaimer: The opinions expressed above are mine, and not endorsed by Honeywell Bull. ------------------------------ Date: 7 Jul 87 08:41:33 edt From: Walter Hamscher Subject: Why AI is not a science Date: Fri, 3 Jul 87 07:29:41 pdt From: norman%ics@sdcsvax.ucsd.edu (Donald A. Norman) I started out writing a message that said this message was 97% true, but that there was an arguable 3%, namely: The problem is that most folks in AI think they are scientists * * * I was going to pick a nit with the word "most". Then, I remembered that the AAAI-86 Proceedings were split into a "Science" track and an "Engineering" track, the former being about half again as thick as the latter... ------------------------------ Date: 8 Jul 87 01:37:17 GMT From: munnari!goanna.oz!jlc@uunet.UU.NET (J.L Cybulski) Subject: Re: Why AI is not a science Don Norman says that AI is not a Science! Is Mathematics a science or is it not? No experiments, no comparisons, thus they are not Sciences! Perhaps both AI and Maths are Arts, ie. creative disciplines. Both adhere to their own rigour and methods. Both talk about hypothetical worlds. Both are used by researchers from other disciplines as tools, Maths is used to formally describe natural phenomena, AI is used to construct computable models of these phenomena. So, where is the problem? Hmmm, I think some of the AI researchers wander into the areas of their incompetence and they impose their quasi-theories on the specialists from other scientific domains. Some of those quasi-theories are later reworked and adopted by the same specialists. Is it, then, good or bad? It seems that lack of scientific constraints may be helpful in advancing knowledge about the principles of science, it seems that the greatest breakthroughs in Science come from those who were regarded as unorthodox in their methods. May be AI is such unorthodox Science, or perhaps an Art. Let us keep AI this way! Jacob L. Cybulski ------------------------------ Date: 07-Jul-1987 0829 From: billmers%aiag.DEC@decwrl.dec.com (Meyer Billmers, AI Applications Group) Subject: Re: AIList Digest V5 #171 Don Norman writes that "AI will contribute to the A, but will not contribute to the I unless and until it becomes a science...". Alas, since physics is a science and mathematics is not one, I guess the latter cannot help contribute to the former unless and until mathematicians develop an appreciation for the experimental methods of science. Ironic that throughout history mathematics has been called the queen of sciences (except, of course, by Prof. Norman). Indeed, physics is a case in point. There are experimental physicists, but there are also theoretical ones who formulate, posulate and hypothesize about things they cannot measure or observe. Are these men not scientists? And there are those who observe and measure that which has no theoretical foundation (astrologists hypothesize about people's fortunes; would any amount of experimentation turn astrology into a science?). I believe the mix between theoretical underpinnings and scientific method makes for science. The line is not hard and fast. By my definition, AI has the right attributes to make it a science. There are theoretically underpinnings in several domains (cognitive science, theory of computation, information theory, neurobiology...) and yes, even an experimental nature. Researchers postulate theories (of representation, of implementation) but virtually every Ph.D. thesis also builds a working program to test the theory. If AI researchers seem to be weak in the disciplines of the scientific method I submit it is because the phenomena they are trying to understand are far more complex and elusive of definition that that of most science. This is not a reason to deny AI the title of science, but rather a reason to increase our efforts to understand the field. With this understanding will come an increasingly visible scientific discipline. ------------------------------ Date: Mon, 6 Jul 87 17:19:38 PDT From: cottrell%ics@sdcsvax.ucsd.edu (Gary Cottrell) Subject: Re: thinking about thinking not being science In article <8707030236.AA29872@flash.bellcore.com> amsler@FLASH.BELLCORE.COM (Robert Amsler) writes: >I think Don Norman's argument is true for cognitive psychologists, >but may not be true for AI researchers. The reason is that the two >groups seek different answers. [....] Speculating about flight might >lead to building other types of aircraft (as certainly those now >humorous old films of early aviation experiments show), but it would >certainly be a bad procedure to follow to understand birds and how >they fly. In fact, the Wright Brothers spent quite a bit of time studying how birds fly, and as a recent Scientific American notes, we may still have a lot to learn from natural systems. A piece of Dennis Conner's boat was based on a whale's tailfin. I think Don's point was that many times AI researchers spend a lot of time theorizing about how humans work, and then use that as justification for their designs for AI systems, without ever consulting the facts. It is certainly true that Cognitive Scientists and AI researchers are at different ends of a spectrum (from NI (Natural Intelligence) to AI), but it would be foolish for AI researchers not to take hints from the best example of an intelligient being we have. On the other hand, it is not appropriate for a medical expert system to make the same mistakes doctors do - sometimes a criterion for a "good" cognitive model. gary cottrell Institute for Cognitive Science C-015 UCSD, La Jolla, Ca. 92093 cottrell@nprdc.arpa (ARPA) (or perhaps cottrell%ics@cs.ucsd.edu) {ucbvax,decvax,akgua,dcdwest}!sdcsvax!sdics!cottrell (USENET) ********************************************************************** THE FUTURE'S SO BRIGHT I GOTTA WEAR SHADES - Timbuk 3 ********************************************************************** ------------------------------ End of AIList Digest ******************** 9-Jul-87 19:43:47-PDT,16476;000000000001 Mail-From: LAWS created at 8-Jul-87 17:37:45 Date: Wed 8 Jul 1987 17:35-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #173 To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 9 Jul 1987 Volume 5 : Issue 173 Today's Topics: Humor - Symbol Grounding References, Theory - Fuzzy Categories, Policy - Symbol-Grounding Metadiscussion ---------------------------------------------------------------------- Date: 7-JUL-1987 15:50:42 From: UBACW59%cu.bbk.ac.uk@Cs.Ucl.AC.UK Subject: References Required. Does anyone have any pointers to the "symbol grounding problem" or some such area? Searches in the literature have proved fruitless. The Joka. ------------------------------ Date: 7 Jul 1987 11:00-EDT From: Spencer.Star@h.cs.cmu.edu Subject: Re: AIList Digest V5 #169 > ...a penguin is not a bird of degree... The point of view that a bird IS a bird, and a rose IS a rose, has limited usefulness. If the question that we are trying to answer is seen as how a person will classify a penguin after having seen one for the first time, I think the answer is clear. A large number of people would not classify a penguin as a bird. A program would likely be more successful at imitating a human response if it based its response on the features of the human answering the query as well as the features of the concept it was trying to recognize. Whether a penguin is a bird then becomes quite dependent on context as well as a simple relation between features and classes. ------------------------------ Date: 8 Jul 87 16:08:27 GMT From: sunybcs!dmark@ames.arpa (David M. Mark) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <974@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: > > >In Article 185 of comp.cog-eng sher@rochester.arpa (David Sher) of U of >Rochester, CS Dept, Rochester, NY responded as follows to my claim that >"Most of our object categories are indeed all-or-none, not graded. A penguin >is not a bird as a matter of degree. It's a bird, period." -- > >> Personally I have trouble imagining how to test such a claim... > >Try sampling concrete nouns in a dictionary. Well, a dictionary may not always be a good authority fro this sort of thing. Last semester I led a graduate Geography seminar on the topic: "What is a map?" If you check out dictionaries, the definitions seem unambiguous, non-fuzzy, concrete. Even the question may seem foolish, since "map" probably is a "basic-level" object/concept. However, we conducted a number of experiments and found many ambiguous stimuli near the boundary of the concept "map". Air photos and satellite images are an excellent example: they fit the dictionary definition, and some people feel very strongly that they *are* maps, others sharply reject that claim, etc. Museum floor plans, topographic cross-profiles, digital cartographic data files on tape, verbal driving directions for navigation, etc., are just some examples of the ambiguous ("fuzzy"?) boundary of the concept to which the English word "map" correctly applies. I strongly suspect that "map" is not unique in this regard! ------------------------------ Date: Mon 6 Jul 87 16:18:12-PDT From: PAT Subject: Re: AIList Digest V5 #170 Talk about walking into a minefield, but here goes. Concerning the Harnad grounding problem. This is lovely stuff, and I save every word for later reading, but it does seem recently to have gone from interesting discussions and arguments to a rather repetitive grinding over the main points again and again. THe result is that Stevan is reduced to repeating himself and reiterating his points in the face of what must seem to him to be increasing stubbornness. I seem to be seeing more and more phrases like '..as I have emphasised earlier..'. All of us who teach are familiar with the syndrome where the 35th occurrence of the same error makes us more exasperated than the first one did. Let me suggest that perhaps nothing much new is being said in these discussions any more, and certainly no-one is saying anything which is going to cause Stevan to change any of his positions. Perhaps the right thing to do is for people to send their comments directly to Harnad, and for him to send us the selections which HE considers worth public airing, together with his responses. That way we will be spared reading all this stuff which is, apparently, of such low intellectual caliber, and Laws will have an easier time, and public feelings will not get to the point which produces letters like David Harwood's. Just an idea. Pat Hayes ------------------------------ Date: Mon, 6 Jul 87 16:56:06 PDT From: cottrell%ics@sdcsvax.ucsd.edu (Gary Cottrell) Subject: Automatic newsgroup creation to reduce aggravation How about some software to automatically create newsgroups after a certain amount of traffic with the same subject line? And an appropriate expiration of the newsgroup after traffic dies down? Then people could decide to add the newsgroup or not. E.g., comp.ai.symbol.grounding.. It doesn't even sound hard enough to be called AI! I am a net.news.software.innocent, however. gary cottrell ------------------------------ Date: Mon, 06 Jul 87 22:03:51 EST From: Tim Daciuk Subject: Symbol Grounding Problem Having read the recent "discussion" regarding the Symbol Grounding Problem, I would have to suggest that I tend to agree with Mr. Harwood. Though the discussion which has taken place on this subject was interesting, it has become, at least to me, tedious and boring. In addition, I think that any- one joining AI-List at this point would find this topic almost impossible to follow, due to the number of references to previous editions of the journal, and due to the highly interactive mode which this discussion has assumed. I do not think that a separate discussion should be started, however, I would suggest that future Symbol Grounding Problem entries be sorted to the bottom of the list. This would allow the list to continue in publishing this important part of AI, and would allow those of us who no longer have the stamina to ponder the implications of blue, green, blue- green, etc., to quit at an appropriate time. Would sorting the list with Symbol Grounding coming at the bottom be very difficult Ken? Tim Daciuk Ryerson Polytechnical Institute Toronto Ontario Canada [That's essentially what I've been doing, although lengthy conference announcements sometimes get sorted even lower. I was holding all of the symbol grounding discussion for the weekend, although that did create some synchronization problems between messages sent to my Arpanet mailbox and replies that went directly to Usenet. I have usually published symbol grounding issues in separate digest issues, making them easier to skip (or save). Usenet readers don't get the benefit of that sorting, of course (but make up for it by eliminating the digesting delay). Sorting to the true "bottom" of an infinite discussion stream would seem a little extreme. -- KIL] ------------------------------ Date: 6 Jul 87 21:38:29 GMT From: harwood@cvl.umd.edu (David Harwood) Subject: An apology for being overly sarcastic I want to apologize for being overly sarcastic with Mr. Harnad. Although I consider my complaint about his postings to be justified, I am sorry about my overly-sarcastic manner. For the record, this apology was my own idea, not involving discussion with others. I simply felt fairly guilty about my irritable responses. (Actually, it is only recently that I've had a chance to read this newsgroup; it has been suggested that I read the moderated newsgroup instead - without posting of course!) \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Briefly responding to a posted reply by B.I. Olasov, also to correspondence by email from D. Stampe (for different reasons): In article <1071@bloom-beacon.MIT.EDU> bolasov@aphrodite.UUCP (Benjamin I Olasov) writes: [...] >Some of the most challenging and interesting problems of AI are philosophical >in nature. I frankly don't see why this fact should disturb anyone. > >Perhaps if more of us pursued our theoretical models with comparable rigor >to that with which Mr. Harnad pursues his, the balance of topics represented >on comp.ai might shift ..... As I tried to make clear - supplying fairly clear examples of his posting style - it is definitely not Mr. Harnad's particular philosophy or theoretical proclivity which irritated me - it was his manner of discussion I was complaining about. (Just as others complained about my sarcasm, more than the content of my complaint.) Among other things, for example, I scarcely consider his arguments to be what you say "rigorous." Some of the discussants themselves have complained, albeit politely, about his somewhat idiosyncratic usage of terminology (among other things). So you are mistaken in your suggestion about my complaint. Rigorous use of very complex and abstract concepts is commonplace in many branches of computer science, eg. semantic specification of languages executed by parallel systems. The level of abstraction and rigor is not at all less than in any area of inquiry, including philosophy or cognitive psychology. On the other hand, I fully agree that both philosophy and psychology have very important and relevant contributions to what is called "artificial intelligence," although it seems to me that too much of the purported interdisciplinary discussion is polemical and political rather than really constructive. And I would add that much, even most, of AI's recent "advance" has been nonsensical propaganda for funding, and devoid of theoretical foundation. Also, I would add that Mr. Harnad - what is clear by his postings - is perhaps only superficially familiar with what are real advances in symbolical "AI", eg development of very powerful systems for automatic deduction, which have practical importance for all of "AI" as well as have rigorous foundations. These surely are not entirely founded on theories of human psychology or on speculative philosophy, and probably should not be, since we would like to consider computing machines which do some things according to specification, and better than we do. I realize very well that some areas of AI are very much harder than others - computer vision comes to mind ;-) and it is obvious to everyone concerned that we need both numerical and symbolical algorithms and representations. (I will not get involved in discussing what S.H. might mean by "symbolical", "analog", "invertible", and so forth - I don't really know.) I think it is also apparent that we might have yet to consider some "connectionist" architectures and algorithms, which perhaps do not admit any simple formal specification of input/output relations. This would invite some philosophical speculation about the adequacy of purely logical specification for development of artificial intelligence. Conversely, we may already have sufficient theoretical basis for 'creating' human-like artificial intelligence, by functional simulation of neurons, although we do have the technology (and moral sense I hope) to de-engineer a human brain. This will surely happen in the distant future only depending on our technology and not on major improvements in our theoretical understanding of neurons. The situation might well be that we can recreate human intelligence which we still largely cannot comprehend by formal specification. In part, these means that psychology, theoretical "AI", even S.H.'s "Total Turing Test" are loose ends as much as interdependent. (As a religious person, I wonder about what this might mean - I recall that an ancient interpretation of the Genesis story said that when mankind ate of the fruit of the knowledge of good and evil - just as the serpent claimed - mankind became endowed with a power like that of God - that is, having the power to create and to destroy worlds. In our times, our technology has surpassed our moral sensibilty - which many computer sceientists say does not exist anyway. Of course, other Jewish tradition has it that many worlds have already been destroyed before this one. I'm not even sure that pursuit of "AI" technology is such a good thing, if it contributes to our destruction or loss of dignity. But who knows, except for God?) Response to a reply by email: \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Message-Id: <8707061548.AA11938@uhmanoa.ICS.HAWAII.EDU> Date: Mon, 6 Jul 87 05:36:04-1000 From: seismo!scubed!sdcsvax!uhccux.UHCC.HAWAII.EDU!nosc!humu!stampe (David Stampe) To: harwood@cvl.umd.edu (David Harwood) In-Reply-To: harwood@cvl.umd.edu's message of 5 Jul 87 21:48:28 GMT Subject: Re: The symbol grounding problem - please start your own newsgroup Status: R You have now posted four messages to comp.ai containing nothing but rude complaints about another's postings on symbol grounding. They are not required reading, and they don't prevent you from reading or posting on other topics. What you MAY NOT do is disturb the newsgroup with irrelevant and loutish postings like your last four. There are people who care about how University of Maryland employees behave in public. If I were you, I'd consider a public apology. David Stampe, Univ. of Hawaii. \\\\\\\\\\\\ I don't have any desire to prevent S.H. from posting, as I have made clear. You are right that I should apologize for being overly sarcastic. He deserved some of it, but I overdid it. I don't like your mention of my employment here - which might be considered to be a threat, either to my employment or to post things which you dislike, even sarcastic complaints. If you did threaten me like this, you would have misjudged me, also misjudged what would be my reaction. In any case, you are right about the apology being due. -David Harwood ------------------------------ Date: Tue, 7 Jul 87 07:33:49 edt From: dg1v+@andrew.cmu.edu (David Greene) Subject: handling the S.G.P issue While some of the discussion has proven interesting, it is become burdensome to sort through and rather recursive as arguments start focusing on what prior arguments meant... Perhaps a seperate bboard would be more appropriate. At the very least, Ken Laws' suggestion that the arguments (and subject lines) be broken into discrete categories seems to go a long way toward making this disscussion palatable if not worthwhile. Mr. Harnad might want to consider proposing a subject taxonomy prefaced with "SGP". David Greene Carnegie Mellon ------------------------------ Date: Wed, 8 Jul 87 11:02:53 GMT From: Caroline Knight Subject: Debating As a so-far passive reader of the grounding problem debate via AIList Digest I have at last been spurred to action: For the proponents of a theory to be able to understand and discuss the positive, the negative and the intersting aspects of it is a sign of strength. For them to resort to personal name calling is not. However I do have sympathy with those who have now started to put the boot in. Especially with those who are tired of the language which is frequently unclear and suspiciously polysyllabic. A thought for those who honestly believe that an idea is wrong and the holder of it would be better off without it:- 1. It is much easier to change one's mind and throw away useless ideas if one has NOT been pushed to defend them tooth and nail. 2. Few ideas (or accepted theories) are completely correct. One can gain more by simply acknowledging that an idea has flaws than by trying to stretch it until it rips. Of course these anomolies might trigger new ideas. Caroline Knight ------------------------------ End of AIList Digest ******************** 11-Jul-87 22:48:25-PDT,22462;000000000000 Mail-From: LAWS created at 11-Jul-87 22:33:13 Date: Sat 11 Jul 1987 22:31-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #174 To: AIList@STRIPE.SRI.COM AIList Digest Sunday, 12 Jul 1987 Volume 5 : Issue 174 Today's Topics: Seminars - Object-Oriented Databases (IBM) & A Model for Distributed Planning (SU) & Pengi: A Theory of Activity (UCB) & Learning Conjunctive Concepts (SU), Conferences - Concurrent Logic Programming, Open Systems Programming & NEXUS meeting at AAAI & Fifth International Machine Learning Conference & Second International Conference on AI in Engineering ---------------------------------------------------------------------- Date: Wed, 01 Jul 87 13:54:22 PDT From: IBM Almaden Research Center Calendar Subject: Seminars - Object-Oriented Databases (IBM) IBM Almaden Research Center 650 Harry Road San Jose, CA 95120-6099 Excerpts from RESEARCH CALENDAR July 6 - 10, 1987 EFFICIENT SUPPORT FOR DERIVED OBJECTS IN RELATIONAL DATABASE SYSTEMS E. Hanson, University of California at Berkeley Comp. Sci. Sem. Tues., July 7 10:00 A.M. Room: B3-247 Recently, an incremental algorithm known as Algebraic View Maintenance (AVM) was proposed for maintaining materialized copies of views. Another incremental view maintenance algorithm called Rete View Maintenance (RVM) is presented in this talk. RVM is based on the Rete network, a type of discrimination network used to support efficient forward chaining rule interpreters in expert systems shells. RVM is known as a statically optimized view maintenance algorithm because the execution plan for maintaining views is compiled in advance into the Rete network. In contrast, AVM is dynamically optimized since an execution plan for maintaining a view is found after each base relation update that affects the view. A statically optimized variation of AVM is also presented. Using algorithms for view maintenance as a starting point, a collection of methods is proposed to allow other kinds of derived objects to be maintained. These include aggregates, database procedures, and views and procedures containing aggregates. Host: S. Finkelstein EPIDEMIC ALGORITHMS FOR REPLICATED DATABASE MAINTENANCE A. Demers, Xerox Palo Alto Research Center Bay Area Syst. Symp. Fri., July 10 11:15 A.M. Room: Front Aud. When a database is replicated at many sites, maintaining mutual consistency among the sites in the face of updates is a significant problem. This paper describes several randomized algorithms for distributing updates and driving the replicas toward consistency. The algorithms are very simple and require few guarantees from the underlying communication system, yet they ensure that the effect of every update is eventually reflected in all replicas. The cost and performance of the algorithms are tuned by choosing appropriate distributions in the randomization step. The algorithms are closely analogous to epidemics, and the epidemiology literature aids in understanding their behavior. One of the algorithms has been implemented in the Clearinghouse servers of the Xerox Corporate Internet, solving long-standing problems of high traffic and database inconsistency. Host: L.-F. Cabrera OBJECT-ORIENTED INTERFACES TO RELATIONAL DATABASES R. Cattell, SUN Microsystems Bay Area Syst. Symp. Fri., July 10 1:30 P.M. Room: Front Aud. Users of engineering workstations have requirements that traditional database systems often do not address. A number of research projects have recently examined addressing engineering requirements with additional semantics that an "object-oriented" database system can provide over a relational database system. My talk will focus on two topics that have received relatively little attention: (1) object-oriented end-user interfaces to databases exploiting the capabilities of an engineering workstation, and (2) the *performance* that these tools and engineering applications require from a database system, without which additional semantics are useless. Examples will be provided from some of our own database and user interface work combining features of object-oriented and relational database models. Users may graphically view and edit a database schema, view database objects that span multiple relations, browse through databases by pointing with a mouse, and display specialized objects such as documents and images stored in a database. To quantify performance, we have proposed a set of benchmarks that measure the simple object-oriented operations that we believe engineering applications most typically execute. I will discuss the results of performing these benchmarks on several relational database systems, and the implications for database system architecture for engineering applications. Host: L.-F. Cabrera For further information on individual talks, please contact the host listed above. ------------------------------ Date: Fri 3 Jul 87 14:17:35-PDT From: Charlie Koo Subject: Seminar - A Model for Distributed Planning (SU) A Model for Distributed Performance -- Synchronizing Plans among Intelligent Agents via Communication Charles C. Koo July 8, Wednesday 9:00am - 10:00am Room 352 Margaret Jacks Hall In a society where a group of agents cooperate to achieve certain goals, agents perform their tasks based on certain plans. Some tasks may interact with tasks done by other agents. One way to coordinate those tasks is to let a master planner generate a plan to begin with, and distribute tasks to individual agents accordingly. However, there are two difficulties with this approach, given that agents are resource-limited. First, the master planner needs to know all the expertise that each agent has. The amount of knowledge sharply increases with the number of specialties. Second, the centralized planning process takes longer turn-around time than if each agent plans for itself. This causes a lot of computing resources not being utilized. Thus, distributed planning is desirable. In this presentation, I will describe a model for synchronizing and monitoring plans autonomously made by intelligent agents via communication. The model suggests an planning algorithm that allows agents to plan in parallel and then synchronize their plans via a commitment-based communication vehicle. Represenation as well as reasoning issues in the distributed environment will be addressed. Communication plays an integral role in planning for synchronization purposes. The communication vehicle includes a minimal set of protocols that enables the synchronization, a set of communication operators and a set of commitment tracking operators. The tracking operators provide means to monitor the progress of plan execution, to prevent delays, and to modify plans with less effort when delays happen. A deadlock detection scheme will also be described. ------------------------------ Date: Mon, 6 Jul 87 08:48:17 PDT From: teresa@ernie.berkeley.edu (teresa diaz) Subject: Seminar - Pengi: A Theory of Activity (UCB) Special Seminar Phil Agre Artificial Intelligence Laboratory Massachusettes Institute of Technology Pengi: An Implementation of a Theory of Activity 2:00 - 4:00 p.m. Friday, July 10, 1987 1011 Evans Hall AI has typically sought to understand the organized nature of human activity in terms of the making and execu- tion of plans. There can be no doubt that people use plans. But before and beneath any plan-use is a continual process of moment-to-moment improvisation. An improvising agent might use a plan as one of its resources, just as it might use a map, the materials on a kitchen counter, or a string tied round its finger. David Chapman and I have been study- ing the organization of the most common sort of activity, the everyday, ordinary, routine, familiar, practiced, unproblematic activity typified by activities like making breakfast, driving to work, and stuffing envelopes. Our theory describes the central role of improvisation and the inherent orderliness, coherence, and laws of change of improvised activity. The organization of everyday routine activity makes strong suggestions about the organization of the human cognitive architecture. In particular, one can get along remarkably well with a peripheral system much as described by Marr and Ullman and a central system made of combinational logic. Chapman has built a system with such an architecture. Called Pengi, it plays a commercial video game called Pengo, in which a player controls a penguin to defend itself against ornery and unpredictable bees. The game requires both moderately complex tactics and constant attention to opportunities and contingencies. I will out- line our theory of activity, describe the Pengi program, and indicate the directions of ongoing further research. ______________________________________________________________________ This information is also kept in usr/public/seminars. ------------------------------ From: Peter Karp Subject: Seminar - Learning Conjunctive Concepts (SU) [Forwarded from the AFLB list.] David Haussler from UC Santa Cruz will be giving a talk at the GRAIL learning seminar this Thursday 7/9 at the Welch Road Conference room at 1:15. This is Room A1110 in Building A at 701 Welch Road, across from the Stanford Barn. Learning Conjunctive Concepts in Structural Domains David Haussler Department of Computer Science, University of California, Santa Cruz, CA 95064 We study the problem of inductively learning conjunctive concepts from examples on structural domains like the blocks world. This class of concepts is formally defined and it is shown that even when each example (positive or negative) is a two-object scene it is NP-complete to determine if there is any consistent concept in this class. We demonstrate how this result affects the feasibility of Mitchell's version space approach and how it shows that it is unlikely that this class of concepts is polynomially learnable from random examples in the sense of Valiant. On the other hand, we show that for any fixed number of objects per scene this class is polynomially learnable from random examples if (1) we allow a larger hypothesis space, or (2) we answer cetrain types of queries in addition to providing random examples. ------------------------------ Date: Mon, 6 Jul 87 14:43:14 PDT From: Ken Kahn Reply-to: Kahn.pa@Xerox.COM Subject: Conference - Concurrent Logic Programming, Open Systems Programming We are pleased to announce that Xerox PARC with support from AAAI will host a workshop on concurrent logic programming, meta-programming, and open systems programming on September 8 and 9 (the first business days after the Fourth IEEE Symposium on Logic Programming in San Francisco). Participation is by invitation only. The purpose of the workshop is to promote informal scientific interchanges between members of various laboratories doing research centered around concurrent logic programming languages such Guarded Horn Clauses and KL1 at ICOT, Parlog at Imperial College, FCP at Weizmann Institute of Science, and Vulcan at Xerox PARC. Other topics of interest include meta-programming to support programming abstractions and issues related to programming large open distributed systems. The format of the workshop will consist of informal presentations and discussions of work in progress. Presentations given at the Fourth SLP are not to be repeated. There will be several panel discussions on topics such as the different proposals for dataflow synchronization in these languages, the role of meta-programming in supporting abstractions, and why it is that there are several indepenent implementation efforts for different dialects of concurrent logic programming languages (or are they committed choice programming langauges or open systems programming languages?). Live demonstrations of software is encouraged. Available computers include Xerox computers running Xerox Common Lisp, Vaxes under Unix 4.2BSD, Sun 3's, IBM PC's, and Macintoshes (SE and II). We will not be covering participants' transportation or living expenses. Lunches will be provided. We are expecting between 20 and 40 participants. If you are interested in coming, or know someone who might be, please send a letter or electronic message indicating what you would like to talk about or demo to: Kenneth Kahn Xerox PARC 3333 Coyote Hill Road Palo Alto, CA 94304 (415) 494-4390 or ArpaNet: Kahn.pa@Xerox.Com Here's the preliminary list of invitees: Ehud Shapiro, Weizmann Institute Shmuel Klinger, Weizmann Institute Vijay Saraswat, CMU Leon Sterling, Case Western Reserve Keith Clark, Imperial College Steve Gregory, Imperial College Andrew Davison, Imperial College M. Huntbach, Imperial College Mitsuhiro Kishimoto, Fujitsu Y. Takayama, ICOT A. Okumura, ICOT Y. Kimura, ICOT H. Seki, ICOT T. Chikayama, ICOT Kazonuri Ueda, ICOT K. Furukawa, ICOT Fernando Pereira, SRI Tony Kusalik, Univ. of Saskatchewan Leon Alkalaj, UCLA Richard O'Keefe, Quintus Compter Systems Bill Kornfeld Lee Naish, Melbourne University G. Levi, University of Pisa Walter Wilson, IBM M. Maher, IBM Carl Hewitt, MIT Will Clinger, Tektronics Mark Miller, Xerox PARC Danny Bobrow, Xerox PARC Curtis Abbott, Xerox PARC Ken Kahn, Xerox PARC Eric Tribble, Xerox PARC ------------------------------ Date: Wed, 8 Jul 87 21:34:05 CDT From: Dan Cerys Subject: Conference - NEXUS meeting at AAAI I don't recall seeing the following on these lists, but this meeting is probably interesting to those interested in the TI Explorer. Please post if it will appear before July 15. The National Explorer Users' Society will meet during AAAI-87 in Conference Room A of the Conference Center House at Seattle Center in Seattle, Washington on Wednesday, the fifteenth of July from three o'clock until six o'clock. 3:00 Welcome, Introductions, and Organization Rich Acuff, Stanford University 3:20 Explorer II Chuck Corley, Strategic Systems Engineering, TI 3:30 New Customer Support Offerings Phil Campbell, Technical Support Center, TI 3:35 Release 3.0 Summary Joyce Statz, User Interface Branch, TI 3:45 TGC, System Training Jim Mynatt, AI Technical Consultant, TI 3:55 Networking, Namespace, and Generic Network Interface Roger Frech, Networking Branch, TI 4:05 New Compiler Features Merrill Cornish, member, Group Technical Staff, TI 4:15 Future Directions Henry Carr, Explorer Software Development, TI 4:25 Educational Marketing Survey John Alden, Educational Marketing, TI 4:30 Bi-Directional Question and Answer 5:10 Break into groups to talk about Explorer II with Chuck Corley LX/Multiprocessing with Kari Karhi Networking with Roger Frech TI Prolog with Dan Cerys NEXUS, the National Explorer Users' Society, met last year as the Explorer Users' Group at AAAI-86. The purpose of the group is to share technical information about the Explorer. There are no dues or membership fees. Membership is open to all Explorer users. To join, send your name, address, phone number, and net address to either of the following addresses: Rich Acuff Stanford University 251 Medical School Office Building Stanford CA 94305 acuff@sumex-aim.stanford.edu Glenda S. McKinney M/S 2201 Texas Instruments P. O. Box 2909 Austin TX 78769 mckinney%dsg%ti-csl@csnet-relay Conference Room A is on the second floor of the Conference Center House, in the northeast corner. The Conference Center House is across the plaza from the Coliseum. ------------------------------ Date: Fri, 10 Jul 87 11:01:21 EDT From: laird@caen.engin.umich.edu (John Laird) Subject: Conference - Fifth International Machine Learning Conference CALL FOR PAPERS FIFTH INTERNATIONAL CONFERENCE ON MACHINE LEARNING Ann Arbor, Michigan June 12-15, 1988 The Fifth International Conference on Machine Learning will be held at the University of Michigan, Ann Arbor, during June 12-15, 1988. The goal of the conference is to bring together researchers from all areas of machine learning. The conference will have open attendance and registration fees. REVIEW CRITERIA In order to ensure high quality papers, each submission will be reviewed by two members of the program committee and judged on clarity, significance, and originality. The best papers will be published in the proceedings, and their authors will be invited to give a talk on their work or describe it at a poster session. All submissions should contain new work, new results, or major extensions to prior work. Summaries and overviews are discouraged. The ideal paper will present a clear description of the learning task being addressed and the proposed solution to that problem. If the paper describes a running system, it should explain that system's representation of inputs and outputs, its performance component, and its learning methods. It should include a detailed example, as well as relate the work to earlier research. Most important, all papers should include some evaluation of the work in the form of substantive results. Papers are not required to take this form, but authors are strongly encouraged to follow this format. SUBMISSION OF PAPERS Each paper must have a cover sheet with the title, authors' names, primary author's address and telephone number, and an abstract of about 200 words. The cover page should also give three keywords that specify the problem area, general approach, and evaluation criteria. Some examples of each are: PROBLEM AREA GENERAL APPROACH EVALUATION CRITERIA Concept learning Genetic algorithms Empirical evaluation Learning and planning Empirical methods Theoretical analysis Language learning Explanation-based Psychological validity Learning and design Connectionist Machine discovery Analogical reasoning The body of the paper must not exceed 13 double-spaced pages in twelve point font, including figures but excluding references. Authors should send four copies of their papers to: Machine Learning Conference Cognitive Science and Machine Intelligence Laboratory The University of Michigan 904 Monroe Street Ann Arbor, MI 48109-1234 Internet: ml88@csmil.umich.edu The deadline for submission of papers is January 15, 1988. Authors will be notified of acceptance by March 1, 1988. Final camera-ready copies of the papers will be due April 1, 1988. Organizing Committee J. E. Laird (chairman) University of Michigan J. H. Holland, S. L. Lytinen, G. M. Olson University of Michigan J. G. Carbonell, T. M. Mitchell Carnegie-Mellon University P. Langley University of California, Irvine R. S. Michalski University of Illinois ------------------------------ Date: Fri, 10 Jul 87 23:40:17 EDT From: sriram@ATHENA.MIT.EDU Subject: Conference - Second International Conference on AI in Engineering SECOND INTERNATIONAL CONFERENCE ON AI IN ENGINEERING The program agenda for the above conference, which is to be held in Boston from August 3-8, 1987, can be obtained from Ms Sandra Elliott, Computational Mechanics Inst., 25 Bridge Street, Billerica, MA 01821, USA (Tel. No: (617)667 5841). I have a copy of the agenda online. If you are interested in getting a copy, send me mail. Some program highlights: Keynote speaker: Dr. Randy Davis, MIT, USA Banquet speaker: Dr. Mark Stefik, Xerox,USA Invited speakers: Dr. John Gero, Univ. of Sydney, Australia Dr. Jean-Claude Latombe, ITMI, France (Currently atStanford Univ.) Dr. B. Chandrasekaran, OSU, USA Panels on: AI in Mechancial Engineering: The Commerical Reality AI in Electrical Engineering: The Commerical Reality AI in Engineering Design: The Research Issues AI in Engineering: The Engineer's Perspective Over 80 papers dealing with various applications of knowledge-based systems, robotics, and natural language processing will be presented. Sriram@athena.mit.edu ------------------------------ End of AIList Digest ******************** 11-Jul-87 22:55:04-PDT,19384;000000000000 Mail-From: LAWS created at 11-Jul-87 22:46:52 Date: Sat 11 Jul 1987 22:43-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #175 To: AIList@STRIPE.SRI.COM AIList Digest Sunday, 12 Jul 1987 Volume 5 : Issue 175 Today's Topics: Queries - ANIMAL in BASIC & XLisp & Monkey and Bananas Benchmark & Conference on Production Planning and Control & Neural Networks & GLISP, Tools - Real Time Expert Systems, Programming - Software Reuse, Law - Liability in Expert Systems, Expert Systems - Plausible Reasoning ---------------------------------------------------------------------- Date: 9 Jul 87 03:04:14 GMT From: David L. Brauer Subject: ANIMAL in BASIC ??? Somewhere in the darkest reaches of my memory I recall seeing a listing of the game ANIMAL in BASIC. It's that old standby introduction to rule-based reasoning that tries to deduce what animal you have in mind by asking questions like "Does it have feathers?", "Does it have hooves?" etc. The problem is that I described this program to my wife and she now wants to program it on an Apple IIc for her elementary school students. I believe I saw the listing in an "Intro to AI" article in some magazine but I'm not sure. I would prefer not to have to help her program the thing from scratch so any pointers would be greatly appreciated. Thanks, David C. Brauer MilNet: dbrauer@NOSC.Mil ------------------------------ Date: Thu 9 Jul 87 08:51:29-PDT From: BEASLEY@EDWARDS-2060.ARPA Subject: clarification I would like to clarify my request for information about XLISP. The particular version i have is XLISP Experimental Object-oriented Language Version 1.6 by David M. Betz for use on the IBM PC and others. Any information would be greatly appreciated. By the way, i have the article from BYTE magazine. The examples didn't work!!!! Please send the info to beasley@edwards-2060.arpa. joe ------------------------------ Date: Fri, 10 Jul 87 10:20:10 SET From: "Adlassnig, Peter" Subject: Monkey and Bananas Benchmark RE: Inquiry for Production Systems Since we finished our PAMELA (PAttern Matching Expertsystem Language) we are interesting in the Monkeys and Bananas benchmark (NASA MEMO FM7(86-51). I wonder how to obtain the source code. In addition to that we would be interested in YAPS (Yet Another Production System) running under VAX/UNIX. Is there any information available. I have no direct access to the ARPANET. Please return mails to my friend's email address: adlassni at awiimc11.bitnet my postal address is: Franz Barachini ALCATEL-ELIN Research Center Floridusgasse 50 A-1210 Vienna Austria ------------------------------ Date: 10 Jul 87 13:46:58 GMT From: dhj@aegir.dmt.oz (Dennis Jarvis) Subject: conference on production planning and control In a (not so) recent posting to comp.ai.digest, it was announced that a conference entitled "Expert Systems and the Leading Edge in Production Planning and Control" would be held from May 10-13 in Charleston, South Carolina. I would like to obtain a copy of the proceedings of that conference - any assistance in this regard would be greatly appreciated. ________________________________________________________________________ Dennis Jarvis, CSIRO, PO Box 4, Woodville, S.A. 5011, Australia. UUCP: {decvax,pesnta,vax135}!mulga!aegir.dmt.oz!dhj PHONE: +61 8 268 0156 ARPA: dhj%aegir.dmt.oz!dhj@seismo.arpa CSNET: dhj@aegir.dmt.oz ------------------------------ Date: Fri, 10 Jul 87 11:04:59 +0200 From: mcvax!idefix.laas.fr!helder@seismo.CSS.GOV (Helder Araujo) Subject: Neural Networks I am just starting working on a vision system, for which I am considering several different architectures. I am interested in studying the utilization of a neural network in such a system. My problem is that I am lacking information on neural networks. I would be grateful if anyone could suggest me a bibliography and references on neural networks. As I am not a regular reader of AIlist I would prefer to receive this information directly. My address: mcvax!inria!lasso!magnon!helder I will select the information and put it on AIlist. Helder Araujo LAAS mcvax!inria!lasso!magnon!helder 7, ave. du Colonel-Roche 31077 Toulouse FRANCE [I have forwarded this to the neuron%ti-csl.csnet@relay.cs.net neural-network list. -- KIL] ------------------------------ Date: 10 Jul 87 14:45:41 GMT From: uwmcsd1!leah!itsgw!nysernic!b.nyser.net!weltyc@unix.macc.wisc.ed u (Christopher A. Welty) Subject: Looking for GLISP I am looking for some references to G-LISP, something written by a guy named Novac (sp?) at Stanford. I don't actually need G-LISP, but I would like to see the papers or any other references. Any help would be much appreciated. With enough interest I'll post to the list.. Christopher Welty - Asst. Director, RPI CS Labs weltyc@cs.rpi.edu ...!seismo!rpics!weltyc ------------------------------ Date: Fri, 10 Jul 87 01:22:56 gmt From: Aaron Sloman Subject: Real Time expert systems Hi, I saw your enquiry about real time expert systems. A UK firm called Systems Designers have used our Poplog system to implement a prototype system called RESCU which can control production of detergent at ICI. This was one of the UK Alvey Programme's "community club" projects, i.e. a number of industrial firms potentially able to benefit from the development helped to fund the prototype demonstration system. They were so pleased with the result that the development work is continuing. They used Poplog on a VAX-730 connected to a variety of monitoring devices displays, etc. The system was written in POP-11 extended by a task specific rule language for which they implemented an incremental compiler using the POP-11 compiler-building tools. There have been various relatively short reports on RESCU in newspapers, etc., as well as conference presentations, but I have not seen a full write-up. If you want to know more about RESCU write to: Mike Dulieu, Systems Designers Plc, Pembroke House, Pembroke Broadway Camberley, Surrey, GU15 3XD England Phone +44 276 686200 I hope this information is of some use. Best wishes Aaron Sloman, U of Sussex, School of Cognitive Sciences, Brighton, BN1 9QN, England UUCP: ...mcvax!ukc!cvaxa!aarons ARPANET : aarons%uk.ac.sussex.cvaxa@cs.ucl.ac.uk JANET aarons@cvaxa.sussex.ac.uk PS Robin Popplestone at University of Amherst Mass (pop@edu.umass.cs) is taking over academic distribution of Poplog in USA. He may have some information about RESCU. He'll be at Amherst and SUN stands at AAAI conference. ------------------------------ Date: 9 Jul 87 03:10:00 GMT From: johnson@p.cs.uiuc.edu Subject: Re: Software Reuse (short title) Object-oriented programming languages like Smalltalk provide a great deal of software reuse. There seems to be several reasons for this. One is that the late bound procedure calls (i.e. message sending) provide polymorphism, so it is easier to write generic algorithms. Late binding encourages the use of abstract interfaces, since the interface to an object is the set of messages it accepts. Another reason is that class inheritance lets the programmer take some code that is almost right and convert it without destroying the original, i.e. it permits "programming by difference". These two features combine to encourage the creation of "application frameworks" or "application toolkits", which are sets of objects and, more importantly, interfaces that let the application developer quickly build an application by mixing and matching objects from existing classes. There are a number of ways that an abstract algorithm can be expressed in these languages. An abstract sort or summation algorithm can be built just using a polymorphic procedure. Abstract "process all" and reduction algorithms are provided by inheritance in the Collection class hierarchy of Smalltalk, and a toolkit can be used to describe the abstract design of a browser or editor from a set of abstract data types, a display manager, and a dialog control component (i.e. the Model/View/Controller system). The Smalltalk programming environment also provides tools to help the user find code and to figure out what it does. While these tools (and the language) could stand some improvement, they already provide a lot of what is needed for code reuse. And they don't use A.I! ------------------------------ Date: Fri, 10 Jul 87 07:53:43 PDT From: George Cross Subject: Re: Liability in Expert Systems Hi, I don't know about any pending cases, but readers interested in this subject should check the article by Christopher J. Gill, High Technology Law Journal, Vol 1, #2, P483-520, Fall 1986 entitled "Medical Expert Systems: Grappling with Issues of Liability." An important legal issue is whether the use of a medical expert system constitutes a product or a service. If an expert system is a product, strict liability applies whereas if it a service then a negligence standard applies. Perhaps some lawyer reading Risks or AILIST could read this article and summarize it for us. It is not easy going. ---- George - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - George R. Cross cross@cs1.wsu.edu Computer Science Department ...!ucbvax!ucdavis!egg-id!ui3!wsucshp!cs1!cross Washington State University faccross@wsuvm1.BITNET Pullman, WA 99164-1210 Phone: 509-335-6319 or 509-335-6636 ------------------------------ Date: Wed, 8 Jul 87 09:56 EDT From: DON%atc.bendix.com@RELAY.CS.NET Subject: Plausibility reasoning >From: Jenny >Subject: so what about plausible reasoning ? >As I read articles on plausible reasoning in expert systems, I come to the >conclusion that experts themselves do not exactly work with numbers as they >solve problems. You are correct in several senses. One, the psychology literature has shown time and time again that human belief revision does not conform to Bayesian evidence accumulation (e.g., Edwards, 1968; Fischhoff & Beyth-Marom, 1983; Robinson & Hastie, 1985; Schum, Du Charme, & DePitts, 1973; Slovic & Lichtenstein, 1971). Two, it does not appear that humans literally use any of the methods. However, the humans do appear to be weighing alternatives. Although, for a period, it may seem that the humans are performing sequential hypothesis testing, for stochastic domains with non-trivial uncertainty, humans gather support for a large set of hypotheses at the same time. They may appear to only gather support for their "favorite"; however, if asked for an ordering over the alternatives or if asked how much they believe the alternatives, it is obvious that they have allowed the evidence to change their beliefs about the non-favorite hypotheses (e.g., Robinson & Hastie, 1985). The question becomes, "what are they doing?" For the sake of argument, let's take your assertion and say they are not explicitly manipulating numbers -- it does seem absurd that the automobile mechanic who can't add simple integers without a calculator could possibly perform the complex aggregations necessary to use numbers. Another possibility is that they are performing a type of non-monotonic logic with the choice of assumptions and generation and testing of possible worlds. This possibility suggests that, if the human is not using numbers at any level, the human's choice of one assumption over another uses a simple set of context sensitive rules. The only time the human should change assumptions (generate an alternative path or possible world) is if the current assumptions are defeated or if some magical attentional process causes the human to arbitrarily try another path. When choosing another path, there should be a fixed set of rules guiding the choice of alternative -- there can be no idea of "this looks a little stronger than that" because such comparisons require a comparison metric which is not built into non-monotonic logics. The psychological research on human search strategies (especially for games such as chess) suggests that humans often abandon one search path to test another which looks like it might be as strong or stronger and then return to the original path. This return to the original path leads to a rejection of the hypothesis that humans maintain a set of assumptions until evidence refutes those assumptions. By my previous argument, then, if non-monotonic logics model human decision making, the humans must be choosing to change path generation based on an attentional mechanism. If numbers are not involved, then the attentional mechanism is probably rule-driven. Of course, I've laid out a straw man. I've said it's either numbers or rules; however, there are probably many other possibilities. The most likely possibility is an analog process something akin to comparisons of weights. If we were to model this process in a computer, we would use numbers; so, we're back to numbers. The trouble with just using numbers, of course, is determining how to combine them under different circumstances and how to interpret them. Plausibility reasoning has been used because it, at least, suggests methods for both of these processes. Something, even an approximation, which has validity at some level, is better than nothing. Rather than turn this into a thesis, let's go on to your next point. >And many of them are not willing to commit themselves into >specifying a figure to signify their belief in a rule. Hum, this sounds like something from Buchanan and Shortliffe. Let's think about the implications of this argument. You're saying, if humans find it difficult to generate numbers to represent their degrees of belief, then numbers must be ineffective. Perhaps even at a higher-level, if humans find some piece of knowledge or knowledge artifact difficult to specify, then it probably is ineffective. What evidence do we have for these claims? What are the implications of these claims? From a personal standpoint, I find any knowledge, beyond the trivial, is difficult to specify in some external formalism (including writing, rules, and probabilities). It seems unlikely that we will ever generate external formalisms which allow painless knowledge transfer. Does that imply that knowledge transfer is hopeless? Let's hope not because that is the modus operandi of the human species. Granted, it will not be perfect, it will be painfull, it will take time, but does that imply that it is worthless? We "know" that human experts have knowledge which is effective. There is growing evidence that purely logical formalisms for representing this knowledge will not work for all problem domains due to the stochastic nature of the domains or the incomplete understanding of the domain. Does this mean that automated problem solving must be limited to non-stochastic domains in which there is a full and complete understanding of the causal relations and elements? I fear that I have left the primary argument which I wanted to use in response to your statement. I looked at statements such as these and asked myself whether "comfort" was a legitimate metric for determining the effectiveness of knowledge. This question suggested an experiment in which different sets of experts were asked to generate the comfortable MYCIN confidence factors, the uncomfortable but definable conditional and a priori probabilities needed for Bayes' theorem, and the interesting, but perhaps not well-defined, probability bounds for the typical Dempster-Shafer formulation. I ran this experiment in which the experts were matched for knowledge in the domain. Each expert was asked to provide the parameters needed for only one of the plausibility reasoning formulisms. The results were that, at a superficial level, humans can provide better MYCIN and Dempster-Shafer parameters than Bayesian numbers. However, when considering how these numbers are used and how errors in the numbers propagate through repeated applications of the aggregation formulae, the Bayesian parameters led to more effective automated decision making than the MYCIN parameters. The performance of the Demspter-Shafer parameters was not significantly better or worse than either system in this test. (This research is documented in two papers -- ask me for references.) The conclusion: the domain expert's comfort is not a legitimate determinant of knowledge effectiveness. >If one obtains two conclusions with numbers indicating some significance, >say 75 % and 80 %, can one say that the conclusion with 80% significance is >the correct conclusion and ignore the other one ? There is a fundamental problem here. If you are refering to percentages, then the numbers cannot add to more than 100. You are correct in that a decision theory for plausibility reasoning must take into account the accuracy of the parameters, and I believe that some researchers have not considered this problem; however, most plausibility reasoning researchers consider the decision theory to be an important component which must be given strict attention. >These numbers do not seem to mean much since they are just beliefs or >probabilties. I alluded to this problem earlier. Actually, if they are probabilities, they mean a lot. Probabilities have clear operational and theoretical definitions. Some, for example Shafer (1981), have suggested that the definition of probabilities can be extended to better account for the subjective nature of the probabilities used in most decision support systems. The real problem is with the MYCIN style confidence factors. Although Heckman (1986) has developed a formal interpretation of confidence factors, the interpretation is ad hoc and it seems difficult to imagine that domain experts use this interpretation. The meaningfulness of the numbers is an important criterion for determining the successful application of the numbers and is one of the strongest arguments for using probabilities and perhaps for using Bayes' theorem. Donald H. Mitchell Don@atc.bendix.com Bendix Aero. Tech. Ctr. Don%atc.bendix.com@relay.cs.net 9140 Old Annapolis Rd. (301)964-4156 Columbia, MD 21045 ------------------------------ End of AIList Digest ******************** 11-Jul-87 22:59:01-PDT,15211;000000000000 Mail-From: LAWS created at 11-Jul-87 22:57:01 Date: Sat 11 Jul 1987 22:53-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #176 To: AIList@STRIPE.SRI.COM AIList Digest Sunday, 12 Jul 1987 Volume 5 : Issue 176 Today's Topics: Binding - Interactive Fiction List, Philosophy of Science - Is AI a Science ---------------------------------------------------------------------- Date: 8 Jul 87 20:16:34 GMT From: engst@tcgould.tn.cornell.edu (Adam C. Engst) Subject: Re: Interactive fiction For those of you who cannot (or don't want to) read the Usenet or Bitnet discussion groups on interactive fiction we are back in mailing list form. If you want to send mail to the list, the address is . . . . . . . . . . >>>> gamemasters@parcvax.xerox.com <<<< Just include "Interactive fiction" on the Subject line so the moderator can separate it out from the adventure game discussion messages. If you want to add yourself to the mailing list (so you get digests every day or so) send a request to . . . . . . . . >>>> gamemasters-request@parcvax.xerox.com <<<< and ask to be added. You can also ask to be deleted from the list, ask for archived mail, or report a mailer failure at the request address. I will be sending the messages that come from Bitnet and Usenet as well, so everyone will have access to all the messages. If anyone has any questions, just email me at either of the below addresses and I'll try to help. Thanks a lot for the discussion up to now and I hope that it will improve even more with the increased audience. Adam C. Engst engst@tcgould.tn.cornell.edu pv9y@cornella.bitnet ------------------------------ Date: 9 Jul 87 14:37 PDT From: Tony Wilkie /DAC/ Subject: Is AI a Science? A Pragmatic Test Offered! I'm inclined to belive that Don Norman is right, and that AI is not a science; which is okay, there being a number of perfectly good, self-respecting fields of study out there that are not sciences. Still, its likely that there have been sensitivities offended and a defense is to be anticipated. In lieu of a more respectable and formal argument in defense of AI being a science, I am prepared to steal from William James and proffer a pragmatic test. The rationale is as follows: 1. Grant moneys are issued by various public and private agencies for the support of research in both sciences and non-sciences 2. Issuing agencies are generally authorized to finance projects falling within their scope of study only. 3. These agencies have some criteria for determining what appropriate projects are. THEREFORE: 4. Any projects funded by an agency as a science (e.g. NSF) are science projects reflecting scientific work (except for method or instrumentation projects). The challenge, then, is to find any researcher working on an AI project funded by a science-supportive agency. If only it were all this easy... Tony Wilkie ------------------------------ Date: Fri, 10 Jul 87 10:34:39 n From: Paul Davis Subject: AI, science & Don Norman Briefly - seems to me that most everyone (including DN himself) has missed out on two key points. First, after Searle, there isn't only *one* AI but two (Searle's strong and weak AI): the first is a suitable target of DN's critique since its whole raison d'etre can be summed up in its idea of AI as `cognitive science', ie; that computer science is a way to approach an understanding of what *existing* intelligent systems do and how they do it. However, let us not forget `weak' AI, which makes no such claims - there is no assumption that the products of weak AI function analagously to "real" intelligent systems, only that they are capable of doing X by some means or another. Second, given that `strong' AI *does* claim to have some intimate relation- ship with cognitive science, its worth asking "is there any other way to study the brain/mind ?". Don Norman castigates (probably correctly) AI for not being a science, but he also fails to point out the likely impossibility of any non-AI-stimulated approaches ever coming to terms with the complexity of the brain. AI models are *NOT* testable!! Just imagine that a keen AI worker comes up with an implementation of his/her model of human brain activity, and that this implementation is so good, and so powerful that it saunters through Mr. Harnad's TTT like a knife through butter.... it is vital to see that there is very little information in this result bearing on the question "is this the correct model of the brain ?". The ONLY way to confirm (test) a `strong' AI model is to demonstrate functionally equivalent hardware behaviour, and psychology is a century or more from being able to do this. Norman seems right to castigate AI workers for excessive speculation unsupported by `real experiments', and undoubtedly, if the aim of `strong' AI is ever to succeed, then we *must* know what it is that we are trying to model, but he should also recognize that AI cannot be tested or developed as other sciences simply because it is unique in studying one domain (computers) with the idea of understanding another (the brain). When AI *is* a science, it will be called psychology.. too long.., paul davis EMBL, Heidelberg, FRG bitnet: davis@embl arpa: davis%embl.bitnet@wiscvm.wisc.edu uucp: ...!psuvax1!embl.bitnet!davis ------------------------------ Date: Fri 10 Jul 87 09:43:03-PDT From: Douglas Edwards Subject: Don Norman on AI as nonscience Don Norman assumes that he knows enough about scientific methods to assert that AI doesn't use them. I don't believe that he, or anyone else, has a good general characterization of how science discovers what it discovers. Especially, I don't believe that he has used scientific methods in determining what scientific methods are. Attempts at characterizing the methods of science typically come from intuitive reflection, or from philosophy, not from science. There are some questions we have to make educated guesses at, because scientific answers are not yet available. Norman's attack on AI is vitiated by the same weakness that vitiated Dresher and Hornstein's earlier attack on AI. The critics' characterizations of scientific methods are far *less* firmly grounded than most assertions being made from within the discipline being attacked. Among intuitive and philosophical theories of scientific method--the only kind yet available--a priori reasoning of the type used in AI plays a prominent role. Exactly what relation such a priori reasoning must have to experimental data is very much an open question. My own background is in philosophy. I have gotten involved in AI partly because I believe, on intuitive grounds, that it *is* a science, and that it has a better shot at giving rise to a truly scientific characterization of scientific methods than philosophy, psychology, linguistics, or neuroscience. (I am not saying anything against interdisciplinary cross-fertilization.) I am now trying to work out a logical characterization of hypothesis formation. Douglas D. Edwards EK225 SRI International 333 Ravenswood Ave. Menlo Park CA 94025 (edwards@warbucks.sri.com) (edwards@stripe.sri.com) ------------------------------ Date: 10 Jul 87 18:37:00 GMT From: jbn@glacier.STANFORD.EDU (John B. Nagle) Reply-to: jbn@glacier.UUCP (John B. Nagle) Subject: Re: AIList Digest V5 #171 In article <8707062225.AA18518@brillig.umd.edu> hendler@BRILLIG.UMD.EDU (Jim Hendler) writes: >When I publish work on planning and >claim ``my system makes better choices than planning program's>'' I cannot verify this other than by showing >some examples that my system handles that 's can't. But of >course, there is no way of establishing that couldn't do >examples mine can't and etc. Instead we can end up forming camps of >beliefs (the standard proof methodology in AI) and arguing -- sometimes >for the better, sometimes for the worse. Of course there's a way of "establishing that couldn't do examples mine can't and etc." You have somebody try the same problems on both systems. That's why you need to bring the work up to the point that others can try your software and evaluate your work. Others must repeat your experiments and confirm your results. That's how science is done. I work on planning myself. But I'm not publishing yet. My planning system is connected to a robot and the plans generated are carried out in the physical world. This keeps me honest. I have simple demos running now; the first videotaping session was last month, and I expect to have more interesting demos later this year. Then I'll publish. I'll also distribute the code and the video. So shut up until you can demo. John Nagle ------------------------------ Date: Fri, 10 Jul 87 20:32:07 GMT From: Caroline Knight Subject: AI applications This is sort of growing out from the discussion on whether AI is a science or not, although I'm more concerned with the status of AI applications. Ever since AI applications started to catch on there has been a growing divide between those who build software as some form of experiment (no comment on the degree of scientific method applied) and those who are building software *FOR ACTUAL USE* using techniques associated with AI. Many people try to go about the second as though it were the first. This is not so: an experimental piece of software has every right to be "toy" in all those dimensions which can be shown to be unnecessary for testing the hypotheses. A fancy interface with graphics does not necessarily make this into a usable system. However most pieces of software built to do a job have potential users some of whom can be consulted right from the start. I am not the first person to notice this, I know. See, for instance, Woods' work on human strengths and weakness or Alty and Coombes alternative paradigm for expert systems or Kidd's work on expert systems answering the wrong questions (sorry I haven't the refs to hand - if you want them let me know and I'll dig them out). I think I have a good name for it: complementary intelligence. By this I mean complementary to human intelligence. I am not assuming that the programmed part of the system need been seen as intelligent at all. However this does not mean that it has nothing to do with AI or cognitive psychology: AI can help build up the computer's strengths and define what will be weaknesses for sometime yet. Cog psy can help define what human's strengths and weaknesses are. Somehow we then have to work out how to put this information together to support people doing various tasks. It is currently much easier to produce a usable system if the whole task can be given to a machine the real challenge for complementary intelligence is in how to share tasks between people and computers. All application work benefits from some form of systems analysis or problem definition. This is quite different from describing a system to show off a new theory. It also allows the builder to consider the people issues: Job satisfaction - if the tool doesn't enrich the job how are you going to persuade the users to adopt it?. Efficient sharing of tasks - just because you can automate some part does not mean you should! Redesign of process? I could go on for ages about this. But back to the main point about whether AI is a science or not. AI is a rather fuzzy area to consider as a science. Various sub-parts might well have gained the status. For instance, vision has good criteria to measure the success of a hypothesis against. I suggest that the area that I am calling complementary intelligence consists of both a science and an engineering discipline. It is a science in which experiments such as those of cog psy can be applied. They are hard to make clear cut but so are many others (didn't you ever have a standard classroom physics experiment fail at school?). It is engineering because it must build a product. And if we want to start a new debate off how about whether it is more profitable to apply engineering methods to software production or to consider it an art - I recently saw a film of Picasso painting in front of a camera and I could see more parallels with some of the excellent hackers I've observed than with what I've seen of engineers at work. (This is valid AI stuff rather than just a software engineering issue because it is about how people work and anyone interested in creating the next generation of programmer's assistants must have some views on this subject!). Caroline Knight This is my personal view. Hewlett-Packard Ltd Bristol, UK ------------------------------ Date: 11 Jul 87 04:48:04 GMT From: isis!csm9a!japplega@seismo.CSS.GOV (Joe Applegate) Subject: Re: Why AI is not a science > From jlc@goanna.OZ.AU.UUCP Sat Feb 5 23:28:16 206 > > May be AI is such unorthodox Science, or perhaps an Art. > Let us keep AI this way! I'm not sure there is any maybe about it! AI development, is in my humble opinion, the most creative expression of the programmers art. Any semi- educated fool can code a program... but the creation of a useful, productivity enhancing application or system is far more art than science! The same is more so in AI development, a query and answer style expert system can be coded in basic by a high school hacker... but the true application for AI is in sophisticated applications that employ high quality presentation techniques that eliminate the ambiguities so often present in a text only presentation. One benefit of the advent of the personal computer is the redirection of software product developent away from data driven environment of DP and accounting and towards the presentation style environment of the non-DP professional. Fortunately, most AI development systems are acknowledging this trend by providing graphical interfaces. Art mimics science and the application of science is an art! Joe Applegate - Colorado School of Mines Computing Center {seismo, hplabs}!hao!isis!csm9a!japplega or SYSOP @ M.O.M. AI BBS - (303) 273-3989 - 300/1200/2400 8-N-1 24 hrs. *** UNIX is a philosophy, not an operating system *** *** BUT it is a registered trademark of AT&T, so get off my back *** ------------------------------ End of AIList Digest ******************** 12-Jul-87 21:47:44-PDT,17108;000000000000 Mail-From: LAWS created at 12-Jul-87 21:34:18 Date: Sun 12 Jul 1987 21:30-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #177 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 13 Jul 1987 Volume 5 : Issue 177 Today's Topics: Theory - Symbol Grounding Poll: Yea's ---------------------------------------------------------------------- Date: 9 Jul 87 03:32:45 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Results of Symbol Grounding Poll (1st of 3 parts) In the poll on whether the symbol grounding discussion was useful and worth continuing there were 24 yea's and 37 nays (with some ambiguous ones I have tried to classify non-self-servingly), so the nays have it. As promised, I am posting the results (yea's in part 2 and nays in part 3) and I will abide by the decision. Perhaps I may be allowed a few parting reflections: (1) It is not entirely clear what the motivation of the nays was: ecological/economic considerations about overuse of the airways or reluctance to perform the dozen or so keystrokes per week (or to put in the software filter) that would flush unwanted topic headers. (2) There were distinct signs of the default option "I can't follow it, therefore it makes no sense" running through some of the nays (and indeed some of the discussion itself). This may be a liability of polling as a method of advancing human inquiry. (3) Along with several thoughtful replies, there was unfortunately also some ad hominem abusiveness, both in the poll and in the discussion. This is the ugly side of electronic networks: unmoderated noise from the tail end of the gaussian distribution. It will certainly be a serious obstacle to making the Net the reliable and respectable medium of scholarly communication that I and (I trust) others are hoping it will evolve into. It may turn out that moderated groups, despite the bottle-necking they add -- a slight step backward from the unique potential of electronic nets -- will have to be the direction this evolution takes. (4) I continue to be extremely enthusiastic about and committed to developing the remarkable potential of electronic networks for scholarly communication and the evolution of ideas. I take the present votes to indicate that the current Usenet Newsgroups may not be the place to attempt to start this. (5) Starting a special-interest Newsgroup every time a topic catches on does not seem like the optimal solution. It is also unclear whether even majority lack of interest should prevail over minority interest when all that seems to be at issue is a keystroke. (Not only is there software to screen out unwanted topics, but to filter multiple postings as well. I have been posting to both comp.ai and comp.cog-eng because they each have a relevant nonoverlapping sub-readership. I subscribe to both; my own version of "rn" only displays multiple postings once. Secondary digests like the ailist are another matter, but everyone knows that half or more of it duplicates comp.ai anyway. The general ecology and economy of the airwaves, on the other hand, should perhaps be deliberated at a higher level, by whoever actually pays the piper.) (6) The current majority status of engineers, computer scientists and programmers on the Net also seems to be a constraint on the development of its broader scholarly potential. Although these two disciplines developed the technology and were the first to use it widely, it's now rather as if Guttenberg and a legion of linotype operators were largely determining not just the form but the content of the printed page. The other academic disciplines need *much* greater representation in the intellectual Newsgroups (such as those devoted to biology, language, philosophy, music, etc.) if the Net's scholarly contribution is ever to become serious and lasting; right now these Newsgroups seem only to be outlets for the intellectual hobbies of the two predominant disciplines. This may just be a quirk of initial conditions and a matter of time. I wlll certainly do my best to get the other disciplines involved in this unique and powerful new medium. [N.B.: I am of course in no way deprecating the great value or contribution to knowledge of the two disciplines I mentioned; I just believe that their incidental monopoly over the electronic networks should be benignly dissolved as soon as possible by the entry of the other disciplines that have a hand in the written word, scholarly communication and the advancement of knowledge. The interdisciplinary field of cognitive science happens to be a microcosm of this larger problem of temporary disciplinary imbalance on the Net, and the subfield of artificial intelligence -- though of course legitimately skewed toward computer science -- seems to be showing some of its effects too, especially on foundational topics like the symbol grounding problem.] -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: 10 Jul 87 11:32:00 EST From: "Robert Breaux" Reply-to: "Robert Breaux" Subject: SYMBOL GROUNDING DIES DOWN It occurs to me that the flare up then dying of symbol grounding in the ai list is an evolution not possible until recently. I believe it is good. In the "old days" prior to electronic bulletin boards, this argument would have raged for years, camps divided, universities would have created "schools of thought", and perhaps books written which would not have stood the "test of time" as a classic issue. Now, we can have "face to face", so to speak, discussions early on, resolve the issues which are not "classic" or seminal, and get on with it. It's GREAT, wouldn't you say? ------------------------------ Date: 9 Jul 87 03:41:27 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Results of Symbol Grounding Poll: Yea's (2nd of 3 parts) [These are the 24 yea's in response to the poll on whether or not to continue the symbol grounding discussion on comp.ai/comp.cog-eng. I have removed names and addresses because I had not asked for permission to repost them. If you wish to communicate with anyone, specify by number (*and* whether "yea" or "nay") and I will forward it to the author.] ------------------------- 1. I am finding the symbol grounding discussion very interesting and would like it to continue. More generally, the community is better served by having too much information flow than too little. I hope the discussion will continue even if most respondents to your poll disagree. ---------------------- 2. I personally don't feel that it's Harwood's place to make a recommendation such as the one he made (rude or otherwise). If the discussion is germane to the stated purpose(s) of the newsgroup (which it is), and is carried on in an intellectually responsible manner (which it certainly has been), why should it not be allowed to continue? Isn't the solution for those who don't find the topic interesting to simply not read the messages bearing that topic on the subject line? After all, any number of discussions can be carried on concurrently. --------------------- 3. I vote to continue on symbol grounding. And by all means, keep going with your good and interesting work. ------------------- 4. I don't read this discussion anymore. I couldn't find the beginning, and never felt that I really understood what the problem was. However, I have absolutely no objection to the discussion continuing. I presume that the discussants get value out of it. ---------------------- 5. Although I only peruse most of the symbol grounding discussion I think it is well placed in comp.ai and I vote to see it continue. Personally, I do not see why intelligent use of the NET needs to be defended but apparently there is always an 'offended' party. --------------------- 6. [re. ailist] I initially found some of the symbol grounding discussion interesting, but at the moment it is getting in my way, interfering with my work of reviewing what is already too much material in AIList. Perhaps a general solution to "what belongs on AIList" is to put lengthy, continuing discussions which are of a temporary nature in separate issues, each clearly titled so it can be deleted by the recepient at the title level without danger of deleting other AIList topics. [Ken Laws, Ailist's moderator, then replied that he was sorting already] Thanks for the reply. Indeed, you are sorting the material already. Thanks for the reference to the mail scanning program. It, or an enhancement of the one I am using, could fill the bill nicely. Perhaps a one-character appendage to the digest name to indicate the issue pertains exclusively to a continuing lengthy discussion? Then, if desired, a smart mailer could automatically omit or delete them. Just a thought. ------------------------ 7. I would, with the following reservation, vote against splitting off this discussion. It is tangential to some important aspects of AI and discussions of this sort tend to emphasize areas which need further scientific exploration. My reservation, which I have until now contained, is that your contributions do tend to be lengthy, wordy, vague, and full of (sigh) ungrounded symbols. At times they also appear to lack respect for the views of other contributors. If you're looking for a soapbox, please find one that doesn't appear in my mailbox. If you have a point to make, and can do so precisely, concisely, and with an open mind towards the responses you receive and respect for their contributors, please contribute to the AIList. This is offered in the spirit of constructive criticism, and I hope you can accept it as such. ---------------- 8. I think symbol grounding discussion are *very* critical to the AIList and count me as pro-discussion on the AIList. -------------------- 9. Ha! I subscribe to quite a few bulletin boards. The symbol grounding problem is the only discussion topic for which I religiously archive all notes. It's far, FAR more important than 99.9% of the drivel you see on the net. What are your critics suggesting? Free up more slots for dumb jokes and sophomoric opinions about the nature of intelligence? I say, "Right on! Keep the symbol grounding discussion going." If you want to be magnanimous, you might request that the discussion be confined to one bulletin board. It seems to inhabit ai, cog-eng, and language boards, at least, now. If you decide to start your own board, however, please let me know. --------------------- 10. Please continue! Critics who care would notice that (in the ailist version at least) these discussions are usually in a posting on their own, and are thus easily discarded by those uninterested. --------------------- 11. Mark one with thumbs up. ------------------- 12. As per our phone conversation this morning... continue the dialogue. ------------------------- 13. Please continue the very enlightening discussion on symbol grounding in its present arena. And thanks very much for the effort you put into explaining quite carefully what you propose. ---------------------------- 14. I consider the recent discussion on the symbol grounding problem to be very interesting and relevant. Please continue. -------------------------- 15. What I am doing is responding to your poll request. Please continue the discussion of the symbol grounding problem. I have not had time to contribute, but I find the contributions, especially yours, quite valuable. (Your contributions are good, but I also value "bad" contributions, since they are often clear examples of the bad philosophy and epistemology which people inflict on themselves and others.) My vote: continue posting. ------------------- 16. Despite the complaints from McCarthy and Minsky, there does seem to be some benefit of the Symbol Grounding discussion for we lurkers. Sometimes I almost think I understand what the issue is. However, I do find it distracting that essentially the same material is arriving by both comp.ai and comp.cog-eng newsgroups. I don't want to unsubscribe to either, but I'd like to have to see the material only once. Is it possible to move this discussion to just comp.cog-eng, since it seems to be the (weak?) AI community that finds much of this correspondence tiresome? I think if you simply announce your intention to operate on one group, and then make all your submissions there (while monitoring both, of course), the news stream will become a bit easier to cope with for many of us. [See earlier material on filtering multiple postings.] --------------------------- 17. I followed your early discussion in symbol grounding but now skip over it. Maybe its gone on too long? But * as long as Ken Laws [ailist] separates it into its own volumes * (as he has been doing) I can skip it and others can follow it as they wish. If he decides this is too much work for him, I would suggest moving it to a different forum. -------------------------- 18. I find the discussion of symbol grounding useful and worth contuinuing. I vote to continue. ----------------------------------- 19. You get my vote for continuing the discussion. --------------------------- 20. Simple response. I don't participate, but I enjoy the discussion. I'm a novice in this area, and seeing exchanges like this help educate. ----------------------- 21. Yes I find it useful and worth continuing. [Mild ad hominem remarks about a prior rude poster deleted] ----------------------------- 22. My response to your request for a vote: I am emphatically *FOR* keeping discussions such as the symbol grounding discussion *ON* Ailist Digest. Though I don't always read all of them (I'm amazed at your energy and ability to sustain these discussions on "paper") as a philosopher I find discussions such as yours the the most important part of the digest. If people think that AI is just computer science, let them start another list. Laws obviously thinks that these discussions are part of AI and he's right. I think that your policy of initially ignoring the rude remarks made against you was a good one. It is unfortunate that some people lose their manners when they go electronic. ---------------------- 23. I vote that you continue the symbol grounding discussion and related topics in the present forum. I've found these articles to be far more enlightening, useful, and relevant than the typical requests and responses for the latest references on KB techniques or expert systems marketing. Not to say that such articles are inappropriate, but that this forum is for all AI-related discussion. Please continue to ignore Booth and Harwood. ------------------------- 24. A difficult question. The discussion HAS been going on at considerable length, but it evolves, and maintains a certain interest. Many people (including me) seem not to work from the same foundation as you, and therefore you need many words to get across what often sounds like reiterations. But if you used fewer words, perhaps we might misunderstand worse than we do. Personally, I think you skirt some important points about categorization, which may be in your book: that it is probably required only for communication (perhaps for a conversation within a single brain, as Gordon Pask would insist); that it usually depends on the existence of a catastrophe function (anywhere near the border of a category, the data may lead unequivocally to more than one result depending on historic and local context); that symbols need not be grounded in real-world phenomena, but in agreed categories constrained by context (people DO communicate about religion and politics, in which fields there is unlikley to be any real-world grounding of the symbols). There are probably other issues. As for continuing the discussion, I would say yes if the contributions could be kept under 75 lines, no otherwise. Or else act as a moderator and submit weekly digests of the arguments people send you privately. ------------------------ -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ End of AIList Digest ******************** 12-Jul-87 22:18:29-PDT,19005;000000000000 Mail-From: LAWS created at 12-Jul-87 22:08:09 Date: Sun 12 Jul 1987 22:05-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #178 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 13 Jul 1987 Volume 5 : Issue 178 Today's Topics: Theory - Symbol Grounding Poll: Nays, Comment - Characteristics of Discussion Lists ---------------------------------------------------------------------- Date: 9 Jul 87 03:44:34 GMT From: mind!harnad@princeton.edu (Stevan Harnad) Subject: Re: Results of Symbol Grounding Poll: Nays (3rd of 3 parts) [These are the 37 nays in response to the poll on whether to continue the symbol grounding discussion in comp.ai/comp.cog-eng. I have removed names and addresses because I had not asked for permission to repost them. If you wish to communicate with anyone, specify by number (*and* whether "yea" or "nay") and I will forward it to the author.] ------- [The first three nays (from Harwood, Minsky and Booth) preceded this poll; enumeration accordingly begins with 4.] ------- 4. Please do not take this personally. I have almost stopped reading comp.ai because of the ridiculous quantity of material being posted by you and Brilliant, and others. This discussion has been completely unuseful to me and I would really like to see it stopped. It is much more like philosophy than AI to me and I am sure there are others who feel the same way but wont tell you. Please stop dominating this newsgroup. ------- 5. Either you start another newsgroup or I unsubscribe to this one. I cannot take any more. ------- 6. Please start your own newsgroup! ------- 7. My vote: *do* start your own news group or private mailing list. This discussion, however interesting it may be to the participants, has gone on too long to continue in "comp.ai". ------- 8. I enjoy skiming your symbol grounding writing though for my research it is totally irrelevant. However since there are relatively few people who do AI who need to consider the TTT (most people in AI are just trying to make machines more intelligent right now) I suspect that the symbol grounding problem better belongs in sci.philosophy.tech. The issues wont come up in the real world for at least 5 years because we are not even close to human emulation at the moment. On the other hand you may be working on psychological modelling. If so then there must be a news group or mailing list more close to that topic than comp.ai. All together I suspect that sci.philosophy.tech is the best place along with periodic notes to comp.ai notifying people that there is a discussion of importance to those who model human beings there. This would get your messages to the relevant people. Also if sci.philosophy.tech doesn't exist for some reason then talk.philosophy would be the next best thing. If the problem is that you can not reach the arpa world that way, I think there is a psychology mailing list. BTW about graded vs ungraded concepts, point taken. On the other hand most of the verbs in any language are regular but most of the verbs used by the speakers of a language are irregular. The dictionary is not meant to be and is not a fair sample of usage. Nor does the set of nouns in the language necessarily correspond to the set of concepts employed by its speakers, (it corresponds to the set of concepts that the speakers find convinient to convey rapidly). However you have presented inconclusive evidence that most concepts are not graded. If you had a dictionary that was sorted by usage and gave the usage of words rather than their definitions you would have better evidence that most concepts are not graded. ------- 9. As to your polling request regarding the symbol grounding issue: I am quite tired of all the traffic it has generated. Considering that no real information has been revealed, I feel it is time to drop it. In the recent time that these postings have filled the newsgroup, most all other worthy postings have vanished. The newsgroup should address a range of pertinent issues that will enlighten subscribers. I feel that the symbol grounding issue has only enlightened me in the use of the 'n' key! While I am on the subject, the cross-posting to 'comp.cog-eng' are atrocious. Either post to one or the other. Most every symbol grounding article has appeared in both. This generated to much traffic on the net and defeats the purpose of making special purpose groups. I thank you for your ability to notice fellow subscribers views. ------- 10. Can't we bag this damn symbol grounding discussion already? If it *must* continue, how about instituting a symbol grounding news group, and freeing the majority of us poor AILIST readers from the burden of flipping past the symbol grounding stuff every morning. ------- 11. I generally do not read the SGP articles simply because I do not understand them (and they are so looong!). If there are a few people interested in reading and discussing SGP, there is no reason to prevent such postings. But if there are also many people who do not want to read that sort of things in comp.ai, then it would be wise to consider the possibility of creating a news-subgroup `comp.ai.sgp'. ------- 12. The ramblings on this topic passed my threshhold of boredom long ago. I'm not proposing censorship, but if you choose to continue the discussion with a smaller group of people who find this topic of interest, I will applaud your good manners. ------- 13. I vote you start your own newsgroup--I was bored with "Symbol Grounding" about 500 kilo-bytes ago. Ditto "The Total Touring Test" or whatever your last filibuster was called. . . . ------- 14. My vote is for ending the discussion on the symbol grounding problem. Thanks. p.s. If you are interested in finding out why I voted against continuing the discussion, please let me know -- I will be glad to oblige. ------- 15. Thank you for taking a poll on whether the symbol grounding problem discussion should or should not continue in comp.ai. My vote is to remove the discussion from this newsgroup. Maybe it could be moved to a new newsgroup talk.symbolgroundingproblem ??? ------- 16. I think that the discussion has been out of hand for a long time now. It doesn't seem to contain any useful insights, and is taking up inordinate resources. Not the least of which is the time spent by the authors expounding their viewpoints. I think that this sort of disagreement is better done in position papers in and letters to journals. The odd use of terms hasn't helped keep the discussion on a high level. Not to point fingers, but your nonstandard use of "analog" made a large number of your posts completely incomprehensible to me until you said that you meant something other than the usual meaning of the term. So, I vote to flush this discussion. ------- 17. Personally, I have been skiping most of the articles in this discussion. I was referred to this newsgroup as a forum for other discussion but have seen little other than what appears to be a war of words from two opposing camps. By now the sides must be set--perhaps it is time to move the discussion from "news" to an e-mail mailing-list. ------- 18. Definitely neither useful nor worth continuing. ------- 19. The manner in which the issue was raised *was* rather rude, but I regret to say that I find much of what was stated about your extended discussions very much to the point. I tried to keep up with discussion; I found it rather interesting at first. But it rapidly became clear that you were all talking at cross purposes, refusing to accept conventional usage or even common-usage-for-the-purpose-of-debate of the key words in question. The appalling level of quotation made things much, much worse and it became well-nigh impossible to ferret out the pearls of insight in the flood of verbiage. I do not wish your discussion to completely vanish from the airwaves, as it were, but without a bit of self-restraint all round, together with some sincere efforts to try to answer one another's objections, I don't think the discussion is particularly useful. (e.g. wrt all-or-none categories: pointing to concrete nouns in the dictionary or to the very special categories that have "hardware support" is not, in my opinion, a sincere effort to meet the objections to the contention that categories are all (or mostly) all-or-none, a rather contrary-to- common-observation position.) Perhaps the new policy on quotation will help: there has been a modest improvement in a couple of the recent postings. I remain hopeful. All I can say is, until things improve quite a bit, I will probably be flushing all the digests with "Symbol Grounding" in the topics list. Sorry. ------- 20. I do not find the symbol grounding problem discussion worthwhile. Thank you for (politely) asking. ------- 21. I vote for discontinuing the discussion. It would be interesting except that there is far too much confusion over who's using what terminology. Probably dozens of articles have been wasted over "well, I don't know what *you* mean by 'analog', but when *I* say 'analog' I mean etc etc etc". ------- 22. You have made an unseemly attempt to bias this vote. The question is not whether your discussion is ``useful and worth continuing,'' but whether we *ALL* need to read or even be sent the truly amazing volume that you seem able to generate on this one topic !?! ** Please remove your discussion from the AI-list (to a new bboard?). ** {And if you find it absolutely necessary to be mad at how stupid and unjust the rest of the world is, go ahead and tally this as a vote for your discussion being useless and not worth continuing} ------- 23. 1. I find it neither interesting nor useful. 2. The arguments, until I stopped following it, somtimeseveral weeks ago, are circular if not repetitive. 3. I've speculated privately that the argruments were cranked out by a machine in someone basement as a Turing Test on the rest of the net. Either that or ... 4. But none of this justifies setting up another news group. comp.ai isn't being used for anything else. For a heavily used group, see comp.sys.ibm.pc. 5. Personally, I'd suggest that you take all of the correspondence. Put it in a folder, and open it again at New Years. Reread it, and write a real paper. ------- 24. Please stop! ------- 25. NO! Please take this discussion to e-mail. It's gone far beyond the point where it's interesting to anyone other than you and the few people still arguing. ------- 26. Stop it! ------- 27. The symbol grounding problem - please start your own newsgroup. DEFINITELY! ------- 28. Although I don't think that AI-list should be strictly limited to discussions of algorithms and similarly down-to-earth items, I do think that the symbol grounding discussion has gotten a bit out of hand and should be conducted privately among the three or four major participants, with perhaps a summary to appear at some future date. ------- 29. In article <977@mind.UUCP> you write: >David Harwood has made two very rude requests (Yes, he was way out of line.) As a former philosophy undergrad and current A.I. grad student, I've found the topic in general to be interesting. BUT . . . I think it should in fact be moved to its own newgroup. Comp.ai is now completely dominated by exchanges between you and Marty Brilliant, Anders Weinstein, etc. After a while, "listening" to a few other people argue gets tedious, no matter how interesting the topic. Frankly, I think people have been frightened away from the newsgroup in the past few months, with the result that there have been no discussions other than this one, unless you count a few requests for info on some language. P.S. I enjoyed your "uncomplemented categories" talk at the Phil/Psych meetings. ------- 30. I vote to cease the endless symbol grounding discussion! ------- 31. I find the discussion neither useful or worth continuing. ------------ 32. Please stop it. I agree with Law that most of the discussion can be carried thru private mail. I can see that R is easier to type than mail ...%....@...... etc but, then use the facilities provided by Unix like aliases etc. I am looking forward to your results. ------- 33. You asked for votes. Mine is... no more on the symbol grounding problem. Thanks for asking. ------- 34. I for one would greatly appreciate having the discussion removed from subsequent AIlists. As in a conference presentation, if a heated topic goes on for too long, the people involved should agree to meet later and discuss the issue amongst themselves without burdening the whole group. You must know by now who the interested parties are; can't you just send mail to each other? ------- 35. It not the discussion per se that I think people object to as much as it is the size of the discussion. The replys are very large, each addressing 15 points of reply to the previous reply. It takes a while to read through the text, and extract some salient points of interest. Having real work to do, I sometimes just file the message, thinking I'll get to it later. I save ALL my mod.ai mail for a time in the near future when I attempt to complete my MS and want to scan back over the current "hot" topics. Unfortunatly I've had to start a special archive just for this discussion, and it's chewing my disk drive all to bits with saved mail. I find the disscussion interesting, and informative but... (Now for the poll): If discussion continues to involve ginormous reply's: END IT If discussion stops taking over whole digests: KEEP IT. ------- 36. I'm sorry but for me the discussion is no longer interesting. ------- 37. I think that this discussion belongs to philosophy, not to AI. I hope that it will relocate itself accordingly. -- Stevan Harnad (609) - 921 7771 {bellcore, psuvax1, seismo, rutgers, packard} !princeton!mind!harnad harnad%mind@princeton.csnet harnad@mind.Princeton.EDU ------------------------------ Date: Thu 9 Jul 87 10:41:46-PDT From: Ken Laws Subject: Characteristics of Discussion Lists [Excerpt from a message to Steven Harnad.] A problem with large, permanent lists is that they are primarily for those on the fringes of a field who want to monitor or join what is happening further in -- but not so far in that it becomes a full-time occupation or involves incomprehensible jargon. The professionals already have channels of communication among themselves (including personal visits, seminars, conferences, publications, and even e-mail or phone calls) and have little time for list discussions that are outside their own exceedingly narrow specialties. As to the suggestion of continuing via e-mail, it's not really so bad. Two options exist. One is to cc everyone on each message, letting the mailer propagate the cc list from one message to another. It is usually easy to add new members to such a discussion, but impossible to drop old ones without retyping the whole list. There is also a problem that BITNET gateways don't add necessary routing information to message that are handed over to the Arpanet. The other option is for one person to maintain a file with all the addresses, headed by a "label:" to suppress the information in the cc field of each message. All traffic is sent to this one individual, who then remails it to the distribution. That's a moderated list. (Anyone can get in this business!) One of the charges in your Nay summary was that discussion of other topics has been down since the fundamentals discussion took over. I believe that's true, although there seems no rational reason for it. Even queries and replies have been reduced, although that could be a coincidence due to the end of the school year and of the proposal year. A few people have dropped off the list because of the volume, many more have added themselves because AIList was beginning to border on their interests. The effects are complex, and certainly not just a linear addition of your text to whatever would have been present anyway. I believe that the proper model of a discussion list is the town meeting. AIList began with my own announcement of myself as moderator, or chairman/speaker of the house. A group of interested individuals formed, and through custom and convention we have worked out an informal social contract that governs the proceedings. Part of the contract is that participants pay reasonable attention to the proceedings, if only to avoid redundant or naive remarks. This, together with the serial nature of current message streams, implies that only one person (more or less ...) has the floor. Part of my job as moderator is to insure a balanced discussion, soliciting (or forwarding) new topics and viewpoints. Not every list is run as a town meeting, but that's my view of AIList. The symbol grounding discussion was carried out with great respect for the participants and with incredible attention to detail. AI needs to grapple with the problems you raised. (Whether AIList needs to is debated in your vote summaries.) The difficulty is simply that people can't pay attention to everything, and your discussion was demanding more attention than they could spare. The other rings of the circus require equal time. Incidentally, much of the personal criticism has been sparked by the one-against-all nature of your discussion. If the level of discussion had been more approachable, we might have had more people joining your cause and providing examples for your position. That would have been more interesting, and might have reached an obvious conclusion or stalemate sooner. It is a common characteristic of net debates, however, that nothing is ever settled. Points that are agreed to are simply dropped, with little or no mention that agreement has been reached, and may even be picked up by some other participant. Net discussions generate a continuous stream of ideas, but conclusions are lacking. I thank you for repeatedly reminding us that conclusions have not been reached in this particular topic area, and hope you will continue to contribute to AIList. -- Ken ------------------------------ End of AIList Digest ******************** 14-Jul-87 22:55:39-PDT,19313;000000000000 Mail-From: LAWS created at 14-Jul-87 22:43:16 Date: Tue 14 Jul 1987 22:35-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #179 To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 15 Jul 1987 Volume 5 : Issue 179 Today's Topics: Review - Spang Robinson 3#6, 6/87 & Spang Robinson 3#7, 7/87 & Canadian AI, 7/87, Report - Strategy Learning with Connectionist Networks, Bibliography - Definitions for Leff a58C & Leff a58C (Part 1 of 2) ---------------------------------------------------------------------- Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: bm654 - Spang Robinson 3#6, 6/87 Summary of Spang Robinson Report on Artificial Intelligence June 1987, Volume 3, No. 6 AI and The Military In 1985, DOD AI activity was 91.1 million with funding of 500 million/year estimated at 1992. The rest discusses summary of military activities, hopes and prospects in the AI field including disillusionment on the part of some in industry. Gary Martins of Intelligent Software is quoted as saying "Early returns from the first two major AI projects under the strategic computing program show few real accomplishments... The autonomous land vehicle projected resulted in the construction of a handsome test track and a huge, lumbering van stuffed with computers running expert systems software. If it travels slowly enough (under three m.p.h), the van is sometimes able to make it all the way around the brightly lit, carefully marked, optically smooth course without serious mishap." "The pilot's associate project aims to produce a refrigerator sized computing system, having functionality comparable to a 3 inch by 5-inch check list car." Charles Anderson of the SDI group said AI would use would be quite low in the SDI project with no increase in the ADI budget for AI applications in spite of the fact that the ADI budget itself is growing." However, the SDI is still spending 200 million per year on AI. Rome Air Force Development Center is building a system to help decide if foreign rocket launches are threats. They also have systems to schedule pilots and aircraft hours. They also have an expert system that links together various office automation tools and can generate its own forms. ()()()()()()()()()()()()()()()()()()()()()()()()()()()()()() Shorts AION corporation's ADS is being extended to CICS and IMS and other IBM data base products. Lockheed has set up a 4.5 million dollar AI center. Symbolics has announced a single chip LISP processor which fits on one card after adding interface and memory chips. Coopers and Lybrand has developed an expert system to monitor brokerage accounts for irregularities. Allan Levine will be manager of Gold Hill's Los Angeles sales office. James McGowan will be Palladian's vice president of sales and marketing nad Thomas Murphy will be their director of sales. 40% of the Japanese Information Processing Association's presentations were related to AI . *(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*(*( This issue also include a directory of people working at various companies, agencies and the like in Military Artificial Intellgience and announcements of various tools and expert systems at the above show. ------------------------------ Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: bm660 - Spang Robinson 3#7, 7/87 Summary of Spang Robinson Report on AI, Volume 3, No. 7, July 1987 The New AI Pioneers: The Knowledge Merchants The market for pre-built expert systems was estimated at 10 to 15 million for 1986 with expected growth to 40 million in 1987. Many developers found extensive customization was needed for each customer while there were many rules that were common to everybody in the application domain. Some info on various expert systems being sold including data on how many sold and time/cost to develop. UNDERWRITER saves three percent in insurance losses while Syntelligence reports a five to ten percent improvement in loss ratios. The numbers on the left are the development cost or times while the numbers on the right are the purchase price. 40 man years: APEX Plan Power (125 sold) ~$34,500 20 man years: APEX Client Profiling ~$100,000 50 man years: Palladian operations planning system ~$100,000 50 man years: Palladian project management system ~$100,000 Sterling Wentworth: PLANMAN, PC based planning system (800 copies, 7500 rules) 8 million: Syntelligence Syntel (risk assesment) ~500,000 Expert Technlogies (yellow page layout) Cogensys: judgement processing for financial service applications (9 systems installed.) ~ $250,000 Composition Systems: publishing systems Eloquent Systems: Hotel Inventory Processing Applicon: circuit design Direct Marketing: Persorft TRansform Logic: Computer Aided Software Engineering (Generates COBOL generation) General Data System, RATER and UNDERWRITER for insurance ~$250,000 _-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_- Real Time Expert Systems on PC's and micros. Texas Instruments developed an expert system in FORTH to control water treatment plan. McDonald Douglas is using a Fuzzy Logic based Forth running on the NOVIX forth engine running 30,000 rules per second. UME Corporation offers an Expert Controller box which is a self-contained controller using expert system technology supporting 5000 rules/second and 16,000 rules total. It is being used in automotive hood stamping process control and for industrial clothes driers. ONSPEC sells a Stand Alone System for $895.00 and Superintendant intended for running Programmable Logic Controllers. The systems support a user-friendly operator interface for the final system, explicit handling of unknown data and retraction of facts. The system handles 1000 rules and 50 rules per second. (A review of this software is in the issue.) ()()()()()()()()()()()()()()()()()()()()()()()()()()()()()() Shorts: Natural Language Incorporated has a product licensing and equity financing agreement with MicroSoft. DATA General will be distributing Gold Hill's products. Teknowledge has named a former Under Secretary of Defense to it's board of directors. Nestor, a maker of a neural-network based system, reported a net loss of $539,252 on revenues of $8,016. MicroProducts is marketing PowerLisp, a virtual memory based system, for Intel 286 and 386 based PC's. Programs in Motion is now offering an expert system with code generators for Pascal, C, dbase III interfacing and form design capabilities. Automated Reasoning is developing expert systems for ATE programming and generates source code in BASIC, C, ATLAS, ADA or Pascal. ------------------------------ Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: bm668 - Canadian AI, 7/87 Summary of Canadian Artificial Intellgience, July 1987, No. 12 Discussion of the Canadian Governments research initiative. The Canadian AI Conference for 1988 will be June 6-10, 1988 in Edmonton. It will be held simultaneously with the Canadian Image Processing and Pattern Recognition Society and Canadian Man-Computer Communications Society meetings. Jose A. Ambros_Ingerson of University of California, Irvine is collecting info re AI applications and efforts in Third World Countries. Canada is setting up a research consortium for AI and robotics. This is similar to MCC and other efforts in that companies produce research that they all can use before competitive additions and applications are made. There was a report on the National Meeting of the Fifth Generation Society. There are a variety of research infrastructures in Canada involving joint industry-academic type efforts. New bindings: Nick Cercone will be Director of the Centre for Systems Science at Simon Fraser University. Randy Goebel is now at the University of Alberta. Brian Schaefer, Beverly Smith, Ian Morrison and Julian Siegel are now at Acquired Intelligence, 2304 Epworth Street, Victoria B. C. V8R 5L2. Report on Research at University of Toronto: Hector Levesque and Ray Reiter are working on formal foundations of knowledge-based systems. John Mylopoulos is working on AI applications to software engineering and databases. Russ Greiner on learning by analogy. Effort to develop an autonomous vision-guided robot. Interpretation of Remotely Sensed Images, e. g. from satellites. Applications include river or lake ice measurements and interpreting weather data for storm forecasting Knowledge Based Debugging system based on MRS. Reviews of "Robotics Research: The Third International Symposium" New Horizons in Educational Computing by Masoud Yazdani The Mathematics of Inheritance Systems by David S. Touretsky Robotics and Ai: An Introduction to Applied Machine Intelligence by Andrew C. Staugaard. Abstracts of papers in Computational Intelligence and some AI Technical Reports. Report on the recent CHI+GI+87 conference on Computer Human Interactions and Graphical Interfaces. ------------------------------ Date: Mon, 13 Jul 87 14:23:53 EDT From: Chuck Anderson Subject: Technical Report: Strategy Learning with Connectionist Networks Strategy Learning with Multilayer Connectionist Representations Chuck Anderson (cwa@gte-labs.csnet) GTE Laboratories Incorporated 40 Sylvan Road Waltham, MA 02254 Abstract Results are presented that demonstrate the learning and fine-tuning of search strategies using connectionist mechanisms. Previous studies of strategy learning within the symbolic, production-rule formalism have not addressed fine-tuning behavior. Here a two-layer connectionist system is presented that develops its search from a weak to a task-specific strategy and fine-tunes its performance. The system is applied to a simulated, real-time, balance-control task. We compare the performance of one-layer and two-layer networks, showing that the ability of the two-layer network to discover new features and thus enhance the original representation is critical to solving the balancing task. (Also appears in the Proceedings of the Fourth International Workshop on Machine Learning, Irvine, June, 1987) ------------------------------ Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: defs for a58C D MAG115 Pattern Recognition\ %V 20\ %N 1\ %D 1987 D MAG116 1985 International Test Conference\ %D 1985 D MAG117 Proceedings IEEE International Symmposium on Circuits and Systems\ %C Kyoto, Japan\ %D JUN 5-7 1985 D MAG118 Proceedings of the Second Australian Conference on Applications of Expe rt Systems\ %C Sydney\ %D 14-16 May 1986 D BOOK66 International Conference on Computers in Engineering Conference and Exh ibit (Las Vegas)\ %D 1984\ %I American Society for Mechanical Engineers D MAG119 Proceedings of the 1986 International Test Conference\ %D SEP 9-11, 1986 D MAG120 1986 IEEE International Conference on Computer Design (Port Chester, NY )\ %D October 6-9, 1986 D BOOK67 1985 Engineering Software IV\ %I Springer Verlag\ %C Berlin-Heidelberg New York\ %D 1985\ %E R. A. Edey D MAG121 American Control Conference (Seattle, WA)\ %D JUN 18-20 1986 D MAG122 1985 Proceedings Annual Reliability and Maintainability Symposium\ %D 1985 D MAG123 Proceedings of the 1986 International Computers and Engineering Confere nce (Chicago, Ill.)\ %D JUL 1986 D MAG124 International Conference on Computer Aided Design (Santa Clara, CA)\ %D 1986 D MAG130 AT&T Technical Journal\ %V 65\ %N 5\ %D SEP-OCT 1986 D MAG131 Pattern Recognition Letters\ %V 5\ %N 3\ %D MAR 1987 D BOOK80 Mathematical Foundations of Computer Science\ %S Lecture Notes in Computer Science\ %V 233\ %I Springer-Verlag\ %C Berlin-New York\ %D 1986 D MAG132 J. Logic Programming\ %V 3\ %N 3\ %D 1986 D BOOK81 GWAI-85 Proceedings of the Ninth German Workshop on Artificial Intellig ence\ %E Herbert Stoyan\ %S Technical Reports on Information Science\ %V 118\ %I Springer-Verlag\ %C Berlin-New York\ %D 1986 D BOOK82 Eighth International Conference on Automated Deduction (Oxford 1986)\ %P 470-488\ %S Lecture Notes in Computer Science\ %V 230\ %I Springer-Verlag\ %C Berlin-New York\ %D 1986 D BOOK83 Algebra, Combinatorics and Logic in Computer Science, Vol I. II, (Gyor, 1983)\ %S Colloq. Math. Soc. Janos Bolyai\ %V 42\ %I North-Holland\ %C Amsterdam-New York\ %D 1986 D BOOK84 Category Theory and Computer Programming (Guildford, 1985)\ %S Lecture Notes in Computer Science\ %V 240\ %I Springer-Verlag\ %C Berlin-New York\ %D 1986 D MAG135 Journal of Logic Programming\ %V 3\ %N 4\ %D 1986\ D MAG136 IEEE Transactions on Geoscience and Remote Sensing\ %V 25\ %N 3\ %D MAY 1987 D MAG137 Soviet Journal of Computer and Systems Sciences\ %V 24\ %N 6\ %D NOV-DEC 1986 ------------------------------ Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: a58C (Part 1 of 2) %A M. J. Amundsen %T The Compact LISP Machine, a Lisp Machine in a Shoe Box %J IEEE National Aerospace and Electronics Conference %V 4 %D 1986 %P 1309-1314 %K H02 %A Robert Buday %T LISP-Machine Maker Symbolics, Spawned at MIT, is Growing Up %J Information Week %N 5 %D MAR 3, 1986 %P 34-37 %K H02 AT16 %A M. Carlsson %T A Microcoded Unifier for LISP Machine Prolog %B Symposium on Logic Programming %D 1985 %P 162-171 %K T02 H02 %A H. Maegawa %T Fast LISP Machine and Lisp Evaluation Processor Eval II-processor Architectur e and Hardware Configuration %J Journal of Information Processing (Japan) %V 8 %N 2 %D 1985 %P 121-126 %K H02 GA01 %A S. Sakamooto %T The Design of a Firmware LISP Machine %R Technology Reports of the Seikei University %I Faculty of Engineering, Fukuoka, Japan %N 41 %D 1986 %P 2751-2752 %K H02 %A H. Schotel %A J. Pijls %T A Prototype From Grammatical Instruction on a LISP Machine %J Informatie (Netherlands) %V 28 %N 1 %D 1986 %P 48-50 %A J. Spoerl %T The Architecture of the Symbolics LISP Machine %J Informatique %V 1 %D 1986 %P 140-144 %A J. M. Switlik %A R. J. Short %T The Database Environment and the LISP Machine %B Artificial Intelligence and Advanced Computer Technology Conference and Exhibition. Proceedings. %D 1986 %A M. Yuhara %T Evaluation of the FACOM Alpha LISP Machine %B Thirteenth Annual International Symposium on Computer Architecture %D 1986 %P 184-190 %K H02 %A V. W. Zue %T The Development of the MIT LISP-Machine Based Research Workstation %J International Conference on Acoustics, Speech and Signal Processing. proceedings %V 1 %D 1986 %P 329-332 %A Y. J. Chao %T Image Processing Methods in Ductile Fracture of Solids %J Mechanics %V 14 %N 1 %D JAN-FEB 1987 %P 57-60 %K AA05 AI06 %A Yu. S. Afonin %T Blocked Branch and Bound Method %J Automation and Remote Control %V 47 %N 8 Part II %D AUG 1986 %P 1107 %K AI03 %A I. B. Muchnik %A P. M. Snegirev %T Algorithm to Estimate the Approximation Accuracy of an Empirical Dependence %J Automation nad Remote Control %V 47 %N 8 Part II %D AUG 1986 %K O06 AI04 O04 %A J. L. Nevins %T Information-Control Aspects of Sensor Systems for Intelligent Robotics %J Journal of Robotic Systems %V 4 %N 2 %D APR 1987 %P 215-228 %K AI07 AI06 %A Hooshang Hemami %A Ralph E. Goddard %T Recognition of Geometrical Shape by a Robotic Probe %J Journal of Robotic Systems %V 4 %N 2 %D APR 1987 %P 237-258 %K AI06 AI07 %A Ren C. Luo %T MIcrocomputer-Based Robot Dynamic Sensing Using Linear Array Sensor for Object Recognition and Manipulation %J Journal of Robotic Systems %V 4 %N 2 %D APR 197 %P 199-214 %K AI06 AI07 H01 %A C. Morandi %A F. Piazza %A R. Capancioni %T Digital Image Registration by Phase Correlation Between Boundary Maps %J IEE Proceedings-E %V 134 %N 2 Part E %P 101-104 %D MAR 1987 %K AI06 %A J. Mantas %T Methodologies in Pattern Recognition and Image Analysis -- A Brief Survey %J MAG115 %P 1-6 %K AI06 %A R. W. Smith %T Computer Processing of Line Images: A Survey %J MAG115 %P 7-16 %K AI06 %A S. J. Roan %A J. K. Aggarwal %A W. N. Martin %T Multiple Resolution Imagery and Texture Analysis %J MAG115 %P 17-34 %K AI06 %A S. Basu %A K. S. Fu %T Image Segmentation by Syntactic Method %J MAG115 %P 35-44 %K AI06 %A Zhen Zhang %A M. Simaan %T A Rule-Based Interpretation System for Segmentation of Seismic Images %J MAG115 %P 45-54 %K AI06 %A Maylor K. Leung %A Yee-Hong Yang %T Human Body Motion Segmentation in a Complex Scene %J MAG115 %P 55-64 %K AI065 %A D. J. Peuquet %A Zhang Ci-Xiang %T An Algoirthm to Determine the Directional Relationship Between Arbitrarily- Shaped Polygons in the Plane %J MAG115 %P 65-74 %K AI06 %A L. G. Shapiro %A R. S. MacDonald %A S. R. Sternberg %T Ordered Structural Shape Matching with Primitive Extraction by Mathematical Morphology %J MAG115 %P 75-90 %K AI06 %A M. R. Korn %A C. R. Dyer %T 3-D Multiview Object Representations for Model-Based Object Recognition %J MAG115 %P 91-104 %K AI06 %A Toshifumi Tsukiyama %A T. S. Huang %T Motion Stereo for Navigation of Autonomous Vehicles in Man-Made Environments %J MAG115 %P 105-114 %K AI06 AA19 %A S. Y. Lee %A S. Yalamanchili %A J. K. Aggarwal %T Parallel Image Normalization on a Mesh Connected Array Processor %J MAG115 %P 115-124 %K AI06 H03 %A H. D. Cheng %A K. S. Fu %T VLSI Architectures for String Matching and Pattern Matching %J MAG115 %P 125-142 %K AI06 O06 H03 %A H. Mellink %A H. Buffart %T Abstract Code Network as a Model of Perceptual Memory %J MAG115 %P 143 %K AI08 %A K. N. Ngan %A A. A. Kassim %A H. S. Singh %T Parallel Image-Processing System Based on the TMS 32010 Digital Signal Processor %J IEE Proceedings E %V 134 %N 2 Part E %D MAR 1987 %K AI06 H03 %A Soundar R. T. Kumara %A R. L. Kashyap %A C. L. Moodie %T Expert System for Industrial Facilities Layout Planning and Analysis %J Computers and Industrial Engineering %V 12 %N 2 %D 1987 %K AA05 AI01 ------------------------------ End of AIList Digest ******************** 14-Jul-87 22:59:05-PDT,23672;000000000000 Mail-From: LAWS created at 14-Jul-87 22:51:34 Date: Tue 14 Jul 1987 22:48-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #180 To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 15 Jul 1987 Volume 5 : Issue 180 Today's Topics: Bibliography - Leff a58C (Part 2 of 2) ---------------------------------------------------------------------- Date: Sat, 11 Jul 1987 17:39 CST From: Leff (Southern Methodist University) Subject: a58C (Part 2 of 2) %A D. Driankov %T An Outline of a Fuzzy Sets Approach to Decison making with Interdependent Goals %J Fuzzy Sets and Systems %V 21 %N 3 %D MAR 1987 %P 275-288 %K O04 AI13 %A J. J. Buckley %T The Fuzzy Mathematics of Finance %J Fuzzy Sets and Systems %V 21 %N 3 %D MAR 1987 %P 257-274 %K AA06 O04 %A S. K. M. Wong %A W. Ziarko %T Comparison of the Probabilistic Approximate Classification and the Fuzzy Set Model %J Fuzzy Sets and Systems %V 21 %N 3 %D MAR 1987 %P 357-362 %K O04 %A W. Karkowkski %A N. O. Mulholland %A T. L. Ward %T A Fuzzy Knowledge Base of an Expert System for Analysis of Manual Lifting Tasks (Case Studies and Applications Contribution) %J Fuzzy Sets and Systems %V 21 %N 3 %D MAR 1987 %P 363 %K AA05 O04 %A S. S. Rao %T Description and Optimum Design of Fuzzy Mechanical Systems %J Journal of Mechanisms, Transmissions and Automation in Design %V 109 %N 1 %D MAR 1987 %K AA05 O04 %P 126-132 %A Heiko Krumm %T Logical Verification of Concurrent Programs %J Angewandte Informatik %N 4 %D APR 1987 %P 131-140 %K AA08 %A Janice I. Glasgow %A Glenn H. MacEwen %T Developing and Proof of a Formal Specification for a Multilevel Secure System %J ACM Transactions on Computer Systems %V 5 %N 2 %D May 1987 %P 151 %K AA08 %A A. Pashtan %T A Prolog Implementation of an Instruction-level Simulator %J Software Practice and Experience %V 17 %N 5 %D MAY 1987 %P 309-318 %K AA08 AA04 T02 %A James L. Flanagan %T Speech Processing an Evolving Technology %J MAG130 %P 2-11 %K AI05 %A James G. Josenhans %A John F. Lynch, Jr. %A Marian R. Rogers %A Richard R. Rosinski %A Wendy P. VanDame %T Speech Processing Application Standards %J MAG130 %P 23-33 %K AI05 %A Robert J. Perdue %A Eugene L. Rissanen %T Conversant 1 Voice System: Architecture and Applications %J MAG130 %P 34-47 %K AI05 %X Conversant is a Registered Trademark %A John G. Ackenhusen %A Syed S. Ali %A James G. Josenhans %A John W. Moffett %A Reuel R. Robertson %A Jaime R. Tormos %T Speech Processing for AT&T Workstations %J MAG130 %P 60-67 %K AI05 %A John G. Ackenhausen %A Syed S. Ali %A David Bishop %A Louis F. Rosa %A Reed Thorkildsen %T Single-Board General-Prupose Speech Recognition System %J MAG130 %P 48-59 %K AI05 %A Martha Birnbaum %A Larry A. Cohen %A Frank X. Welsh %T A Voice Password System for Access Security %J MAG130 %P 68-74 %K AI05 %A Bishnu S. Atal %A Lawrence R. Rabiner %T Speech Research Directions %J MAG130 %P 75-88 %K AI05 %A Knut Conradsen %A Gert Nilsson %T Data Dependent Filters for Edge Enhancement of Landsat Images %J Computer Vision, Graphics, and Image Processing %V 38 %N 2 %D MAY 1987 %P 101-121 %K AI06 %A Ken-Ichi Kanatani %T Structure and Motion from Optical Flow Under Perspective Projection %J Computer Vision, Graphics, and Image Processing %V 38 %N 2 %D MAY 1987 %P 122-146 %K AI06 %A Azriel Rosenfeld %T Picture Processing: 1986 %J Computer Vision, Graphics, and Image Processing %V 38 %N 2 %D MAY 1987 %P 147 %K AI06 %A W. Greblicki %A M. Pawlak %T Necessary and Sufficient Conditions for Bayes Risk Consistency of a Recursive Kermnel Classification %J IEEE Transactions on Information Theory %D MAY 1987 %V 33 %N 3 %P 408-411 %K O04 %A V. Wispfenning %T The Complexity of the Word Problem for Abelian I-Groups %J Theoretical Computer Science %V 48 %N 1 %D 1986 %P 127 %K AI14 AI10 %A A. V. Zhozhikashvili %A V. L. Stefanyuk %T The Category Theory in Problems of Knowledge Representation and Learning %J Soviet Journal of Computer and Systems Sciences %V 24 %N 5 %D SEP-OCT 1986 %P 11-23 %K AI16 AI04 %A Ye. K. Gordiyenko %T Implementation of Search Functions of the FRL Language Using a Two-Tag Associative Memory %J Soviet Journal of Computer and Systems Sciences %V 24 %N 5 %D SEP-OCT 1986 %P 43-58 %K AI03 %A L. I. Feygin %T Estimation of the Value of the Planning Horizon in the Case of Fuzzy Durations of the Operations %J Soviet Journal of Computer and Systems Sciences %V 24 %N 5 %D SEP-OCT 1986 %P 97-101 %K AI09 O04 %A Ronald R. Yager %T On the Dempster-Shafer Framework and New Combination Rules %J Information Sciences %V 41 %N 2 %D MAR 1987 %P 93-138 %K O04 %A J. C. A. Van Der Lubbe %A D. E. Boekee %A Y. Boxma %T Bivariate Certainty and Information Measures %J Information Sciences %V 41 %N 2 %D MAR 1987 %P 139-170 %K O04 %A M. A. Zuenkev %A A. S. Kulguskin %A A. G. Poletykin %T Forming Similarity Relations in Analogy-Driven Systems %J Automation and Remote Control %V 47 %N 11 Part 2 %D NOV 1986 %P 1543-1551 %K AI16 %A S. Daley %A f. F. Gill %T Attitude Control of a Spacecraft Using an Extended Self-Organizing Fuzzy Logic Control %J Proceedings of the Institution of Mechanical Engineers Part C %V 201 %N 2 %D 1987 %P 97-106 %K AA05 O04 %A G. Jumarie %T A Concept of Observed Weighted Entropy and its Application to Pattern Recognition %J MAG131 %P 191-194 %K AI06 %A J. H. Kim %T Distributed Inference for Plausible Classification %J MAG131 %P 195-202 %K AI06 %A J. Ma %A X. Lu %A C. Wu %T A Motion Constraint Equation Under Space-Varying or Time Varying Illumination %J MAG131 %P 203-206 %K AI06 %A M. Werman %A A. Y. Wu %A R. A. Melter %T Recognition and Characterization of Digitized Curves %J MAG131 %P 207-214 %K AI06 %A G. Cristobal %A J. Bescos %A J. Santamaria %A J. Montes %T Wigner Distribution Representation of Digital Images %J MAG131 %P 215-222 %K AI06 %A S. Peleg %A D. Keren %A L. Schweitzer %T Improving Image Resolution Using Subpixel Motion %J MAG131 %P 223-226 %K AI06 %A M. C. Yuan %A J. G. Li %T A Production System for LSI Chip Anatomizing %J MAG131 %P 227-232 %K AI06 %A R. D. Grisell %T Noniterive Correction of Images and Motion Sequences %J MAG131 %P 223-242 %K AI06 %A P. Fua %A A. J. Hanson %T Resegmentation Using Generic Shape: Locating General Cultural Objects %J MAG131 %P 243 %K AI06 %A A. M. Rustamov %A N. G. Dzhanibekova %A V. G. Zakiev %T Development of the Automated System on the Analysis of Reader Demand in Regional Integral Automated Library-Bibliography Systems %J Nauchno-Tekhnicheskaya Informatsiya, Seriya II - Informatsionnye Protsessy I Sistemy %N 3 %D 1987 %P 11-18 %K AA14 %A I. A. Bolshakov %T Pure Automatic Seplling Correction Based on the Keyboard Model of Common Errors %J Nauchno-Tekhnicheskaya Informatsiya, Seriya II - Informatsionnye Protsessy I Sistemy %N 3 %D 1987 %P 11-18 %A K. V. K. K. Prasad %A T. S. Lamba %T Natural Language Interface Based on Keyword Extraction Using AWK %J Microprocessors and Microsystems %V 11 %N 3 %D APR 1987 %K AI02 %P 157-160 %A A. N. Averkin %A V. B. Tarasov %T The Fuzzy Modeling Relation and its Application to Artificial Intelligence %J MAG122 %P 3-24 %K O04 %A A. V. Alexeyev %A A. N. Borisov %A V. I. Glushkov %A O. A. Krumberg %A G. V. Merkuryeva %A V. A. Popov %A N. N. Slyadz %T A Linguistic Approach to Decision-Making Problems %J MAG123 %P 25-42 %K AI02 AI13 O04 %A R. A. Aliev %T Production Control on the Basis of Fuzzy Models %J MAG123 %P 43-56 %K O04 %A A. F. Blishun %T Fuzzy Learning Models in Expert Systems %J MAG123 %P 57-70 %K AI01 AI04 O04 %A V. E. Zhukovin %A F. V. Burshtein %A E. S. Korelov %T A Decisoin Making Model with Vector Fuzzy Preference Relation %J MAG123 %P 71-80 %A S. G. Svarovski %T Usage of Linguistic Variable Concept for Human Operator Modelling %J MAG123 %P 107-114 %K O04 AI02 %A D. A. Pospelov %T Fuzzy Reasoning in Pseudo-Physical Logics %J MAG123 %P 115-120 %K O04 %A S. V. Chesnokov %T The Effect of Semantic Freedom in the Logic of Natural Language %J MAG123 %P 121-154 %K AI02 O04 %A D. I. Shapiro %T Human Specifics, Fuzzy Categories and Counteraction in Decision Making Problems %J MAG123 %P 155-170 %K AI13 O04 %A I. A. Newman %A R. P. Stallard %A M. C. Woodward %T A Hybrid Multiple Processor Garbage Collection Algorithm %J The Computer Journal %V 30 %N 2 %D APR 1987 %P 110-118 %K T01 H03 %A J. L. Dupouey %T Using Artificial Intelligence Languages for the Calculation of Inbreeding Coefficients - New Tools for an Old Problem %J Computers in Biology and Medicine %V 17 %N 2 %D 1987 %P 71-74 %K AA10 %A Rob Gerth %A W. P. de Roever %T Proving Monitors Revisited: a First Step Towards Verifying Object Oriented Systems %J Fund. Inform. %V 9 %D 1986 %N 4 %P 371-399 %K AA08 %A P. T. Cox %T On Determining the Causes of Nonunifiability %J J. Logic Programming %V 4 %D 1987 %N 1 %P 33-58 %K AI14 AI10 %A Peter van Emde Boss %T A Semantical Model for Integration and Modularization of Rules %B BOOK80 %P 78-92 %K AI01 AI16 %A Ken Hirose %T An Approach to Proof Checker %B BOOK80 %P 113-127 %K AA13 AI14 AI11 %A Guy Jumarie %T New Decision Rules in Statistical Pattern Recognition %J Kybernetes %V 16 %D 1987 %N 1 %P 11-18 %K AI06 %A A. V. Kabulov %A B. I. Zufarov %T Logical Methods for the Design of Optimal Correctors of Heuristic Algorithms %B "Fan" %C Tashkent %D 1985 %P 11-17 %K AI16 %A I. V. Kotel'nikov %T An Algorithm for Constructing a Set of Irredundant Fuzzy Sets %J Avtomat. i. Telemekh. %D 1986 %N 9 %P 139-144 %K O04 %A M. A. Nait Abdallah %T Al-Khowarizmi: A Formal System for Higher Order Logic Programming %B BOOK80 %P 545-553 %K AI10 %A Zbigniew W. Ras %A Maria Zemankova %T Learning in Knowledge Based Systems, a Possibilistic Approach %B BOOK80 %P 630-638 %K AI04 O04 %A D. Snyers %T Theorem Proving Techniques and P-Functions for Logic Design and Logic Programming %J Philips J. Res %V 41 %D 1986 %N 5 %P 560-505 %K AA04 AI11 AI10 %A Zbigniew M. Wojcik %T The Rough Sets Utilization for Linguistic Pattern Recognition %J Bull. Polish Acad. Sci. Tech. Sci %V 34 %D 1986 %N 5-6 %P 285-312 %K AI06 AI02 %A S. K. M. Wong %T Algorithm for Inductive Learning %J Bull. Polish Acad. Sci. Tech. Sci. %V 34 %D 1986 %N 5-6 %P 271-276 %K AI04 %A S. K. M. Wong %A Wojciech Ziarko %T Remarks on Attribute Selection Criterion in Inductive Learning Based on Rough Sets %J Bull. Polish. Acad. Sci. Tech. Sci %V 34 %D 1986 %N 5-6 %P 273-283 %K AI04 %A W. Bibel %A Ph. Jorrand %T Fundamentals of Artificial Intelligence. An Advanced Course. %S Lecture Notes in Computer Science %V 232 %I Springer-Verlag %C Berlin-New York %D 1986 %K AI16 AT15 %A V. Arvind %A Somenath Biswas %T An O($N sup 2$) algorithm for the Satisfiability Problem of a Subset of Propositional Sentences in CNF that Includes all Horn Sentences %J Inform. Process. Lett %V 24 %D 1987 %P 67-69 %K O06 AI10 %A Luis Farinas del Cerro %A Martti Pentonnen %T A Note on the Complexity of the Satisfiability of Modal Horn Clauses %J J. Logic Programming %V 4 %D 1987 %N 1 %P 1-10 %K AI11 O06 %A Fracoise Fogelman-Soulie %A Gerard Weisbuch %T Random Iterations of Threshold Networks and Associative Memory %J SIAM J. Comput %V 16 %D 1987 %N 1 %P 203-220 %K AI16 AI08 %A Erik Tiden %T First-order Unification in Combinations of Equational Theories (Ph. D. Thesis) %I Royal Institute of Technology %C Stockholm %D 1986 %K AI14 AI11 %A Moshe Y. Vardi %T Querying Logical Databases %J J. Comput. System Sci %V 33 %D 1986 %N 2 %P 142-160 %K AA09 AI10 %A Zbigniew M. Wojcik %T Contextual Information Research within Sentence with the Aid of the Rough Sets %J Bull. Polish Acad. Sci. Tech. Sci %V 34 %D 1986 %N 5-6 %P 313-330 %K AI02 O04 %A Friedhelm Hinz %T Regular Chain Code Picture Languages of Nonlinear Descriptional Complexity %B BOOK80 %P 414-421 %K AI06 %A Stephen D. Brookes %T A Fully Abstract Semantics and a Proof System for an ALGOL-like language with Sharing %B Mathematical Foundations of Programming Semantics %P 59-100 %S Lecture Notes in Computer Science %I Springer-Verlag %C Berlin-New York %D 1986 %K AA08 %A Susanne Graf %T A Complete Inference System for an Algebra of Regular Acceptance Models %B BOOK80 %P 386-395 %K AI10 %A Laszlo Bela Kovacs %T Automated Protocol Verification %B Kozl.-MTA Szamitastech. Automat. Kutato Int. Budapest %N 33 %D 1985 %P 37-45 %A M. J. Beeson %T Proving Programs and Programming Proofs %B Logic, Methodology and Philosophy of Science, VII %S Stud. Log Foundations Math. %V 114 %I North-Holland %C Amsterdam-New York %D 1986 %K AA08 AI16 %A Anne-Marie Deroualt %A Bernard Merialdo %T Language Modelling Using a Hidden Markov Chain with Application to Automatic Transcription of French Stenotypy %B Semi-Markov Models %I Plenum %C New York-London %D 1986 %K AI02 %A A. J. Baddeley %T Stochastic Geometry and Image Analysis %B Mathematics and Computer Science (Amsterdam 1983) %P 1-18 %S CWI Monographs %V 1 %I North-Holland %C Amsterdam-New York %D 1986 %K AI06 %A A. G. Ivakhenko %A S. A. Petukhova %T Objective Computerized Clustering. I. Theoretical Questions %J Soviet J. Automat. Inform. Sci %V 19 %D 1986 %N 3 %P 1-9 %K O06 %A Hassan Ait-Kaci %T LOGIN: A Logic Programming Language with Built-in Inheritance %J MAG132 %P 185-215 %K AI10 %A Marco Bellia %A Giorgia Levi %T The Relation Between Logic and Functional Languages: A Survey %J MAG132 %P 217-236 %K AT08 %A Karl-Hans Blasius %T Equality Reasoning with Equality Paths %B BOOK81 %P 57-76 %K AI14 %A Wolfram Buttner %T Unification in the Data Structure Sets %B BOOK82 %P 470-488 %K AI14 AA08 %A Ahlenm Ben Cherifs %A Pierre Lescane %T An Actual Implementation of a Procedure that Mechanically Proves Termination of Rewriting Systems Based on Inequalities Between Polynomial Interpretations %B BOOK82 %P 42-51 %K AI14 AI11 %A P. Ciancarini %A P. Degano %T An Approach to Proving Properties of Nonterminating Logic Programs %B BOOK83 %P 223-243 %K AI14 AA08 O02 %A Hubert Comon %T Sufficient Completeness, Term Rewriting Systems and "Anti-Unification" %B BOOK82 %P 128-140 %K AI14 AI11 %A P. Tox Cox %A T. Pietrzykowski %T Causes for Events: Their Computation and Applications %B BOOK82 %K AI11 temporal reasoning %A A. J. J. Dick %A R. J. Cunningham %T Using Narrowing to Do Isolation in Symbolic Equation Solving %B BOOK82 %P 272-280 %K AI14 %A Roland Dietrich %T Relating Resolution and Algebraic Completion for Horn Logic %B BOOK82 %P 62-78 %K AI14 AI10 AI11 %A B. Fronhofer %T On Refinements of the Connection Method %B BOOK83 %P 391-401 %A Isabelle Gnaedig %A Pierre Lescanne %T Proving Termination of Associative Commutative Rewriting Systems by Rewriting %B BOOK82 %P 52-61 %K AI14 AI11 %A Richard Gobel %T Completion of Globally Finite Term Rewriting Systems for Inductive Proofs %B BOOK81 %P 101-110 %K AI11 AI14 %A I. R. Goodman %T Some Asymptotic Results for the Combination of Evidence Problem %J Math. Modelling %V 8 %D 1987 %P 216-221 %K O04 O06 %A Alexander Herold %T Combination of Unification Algorithms %B BOOK82 %P 450-469 %K AI11 AI14 %A Douglas Howe %T Implementing Number Theory: an Experiment with Nuprl. %B BOOK82 %P 404-415 %K AA13 AI11 AI14 %A Tadashi Kanamori %A Hiroshi Fujita %T Formulation of Induction Formulas in Verification of Prolog Programs %B BOOK82 %P 281-299 %K AI14 AI11 O02 %A Deepak Kapur %A Paliath Narendran %A Hantao Zhang %T Proof by Induction Using Test Sets %B BOOK82 %P 99-117 %K AI14 AI11 %A Deepak Kapur %A Paliath Narendran %T NP-Completeness of the Set Unification and Matching Problems %B BOOK82 %P 489-495 %K O06 AI11 %A Thomas Kaufl %T Program Verifier "Tatzelwurm": Reasoning About Systems of Linear Inequalities %B BOOK82 %P 300-305 %K AA13 AA08 AI11 %A Younghwan Lim %T The Heuristics and Experimental Results of a New Hyperparamodulation: HL- Resolution %B BOOK82 %P 240-253 %K AI11 %A Rasiah Loganantharaj %A Robert A. Mueller %T Parallel Theorem Proving with Connection Graphs %B BOOK82 %P 337-352 %K AI11 H03 %A Zohar Manar %A Richard Waldinger %T How to Clear a Block: Plan Formulation in Situational Logic %B BOOK82 %P 622-640 %K AI07 AI09 AI11 %A Ursula Maritn %A Tobias Nipkow %T Unification in Boolean Rings %B BOOK82 %P 506-513 %K AI14 AI11 %A Jalel Mzali %T Matching with Distributivity %B BOOK82 %P 496-502 %K O06 AI11 %A Sanjal Narain %T A Technique for Doing Lazy Evaluation in Logic %J MAG132 %P 259-276 %K AI10 %A Hung T. Nguyen %T On Modeling of Expert Knowledge and Admissibility of Uncertainty Measures %J Math. Modelling %V 8 %D 1987 %P 222-226 %K O04 AI01 %A Hans-Jurgen Ohlbach %T Theory Unification in Abstract Clause Graphs %B BOOK81 %P 77-100 %K AI14 AI11 %A F. Oppacher %A E. Suen %T Controlling Deduction with Proof Condensation and Heuristics %B BOOK82 %P 384-393 %K AI11 AI14 %A Lawrence C. Paulson %T Natural Deduction as Higher-Order Resolution %J MAG131 %P 237-258 %K AI10 AI11 %A David A. Plaisted %T Abstraction Using Generalization Functions %B BOOK82 %P 365-376 %K AI11 %A D. Rydeheard %T A Categorical Unification Algorithm %B BOOK84 %K AI14 AI11 %A Manfred Schmidt-Schauss %T Unification in Many-Sorted Equational Theories %B BOOK82 %P 538-552 %K AI14 AI11 %A Manfred Schmidt-Schauss %T Unification in a Many Sorted Calculus with Declarations %B BOOK81 %P 118-132 %K AI14 AI11 %A Hans-Albert Schneider %T An Improvement of Deduction Plans: Refutation Plans %B BOOK82 %P 377-383 %K AI11 %A O. Stepankova %A P. Stepanek %T And/or Schemes and Logic Programs %B BOOK83 %P 765-776 %K AI10 AI03 %A Mandayam Thathachar %A P. S. Sastry %T Learning Optimal Discriminant Functions Through a Cooperative Game of Automata %J IEEE Trans. Systems Man Cybernet. %V 17 %D 1987 %N 1 %P 73-85 %K AI12 AI04 %A Erik Tiden %T Unification in Combinations of Collapse-Free Theories with Disjoint Sets of Function Symbols %B BOOK82 %P 431-449 %K AI11 AI14 %A F. Winkler %A B. Buchberger %T A Criterion for Eliminating Unnecessary Reductions in the Knuth-Bendix Algorithm %B BOOK83 %P 849-869 %K AI14 AI11 %A L. Wos %A W. McCune %T Negative Paramodulation %B BOOK82 %P 229-239 %K AI14 AI11 %A Martin Abadi %A Zohar Manna %T Modal Theorem Proving %B BOOK82 %P 172-189 %K AI11 %A Peter B. Andrews %T Connections and Higher-Order Logic %B BOOK82 %P 1-4 %K AI11 AI10 %A Leo Bachmair %A Nachum Dershowitz %T Commutation, Transformation, and Termination %B BOOK82 %P 5-20 %K AI11 AI14 %A Julian Besag %T On the Statistical Analysis of Dirty Pictures %J J. Royal Statistical Society Series B %V 48 %D 1986 %N 3 %P 259-302 %K AI06 %A R. Book %T On the Unification Hierarchy %B BOOK81 %P 111-117 %K AI14 AI11 %A Frank Malloy Brown %T A Commonsense Theory of Nonmonotonic Reasoning %B BOOK82 %P 209-228 %K AI15 %A Hans-Jurgen Burckert %T Some Relationships Between Unification, Restricted Unification, and Matching %B BOOK82 %P 514-524 %K AI11 AI14 O06 %A Cynthia Dwork %A Paris Kanellakis %A Larry Stockmeyer %T Parallel Algorithms for Term Matching %B BOOK82 %P 416-430 %K AI11 O06 H03 AI14 %A Norbert Eisenger %T What You Always Wanted to Know About Clause Graph Resolution %B BOOK82 %P 316-336 %K AI11 %A M. Falaschi %A Giorgia Levi %A C. Palamidesi %T The Formal Semantics of Processes and Streams in Logic Programming %B BOOK83 %P 363-378 %K AI10 O02 %A Jieh Hsiang %A Michael Fusinowitch %T A New Method for Establishing Refutational Completeness in Theorem Proving %B BOOK82 %P 141-152 %K AI14 AI11 %A Gerhard Jaeger %T Some Contributions to the Logical Analysis of Circumscription %B BOOK82 %P 154-171 %K AI15 AI11 %A Kurt Konolige %T Resolution and Quantified Epistemic Logics %B BOOK82 %P 199-208 %K AI10 AI11 AI14 %A Xu Hua Liu %T Generalized Resolution Using Paramodulation %J Kexue Tongbao (English Edition) %V 31 %D 1986 %N 21 %P 1441-1444 %K AI11 AI14 %A Neil V. Murray %T Theory Links in Semantic Graphs %B BOOK82 %P 353-364 %K AI16 %A David A. Plaisted %T A Simple Nontermination Test for the Knuth-Bendix Algorithm %B BOOK82 %P 69-88 %K AI11 AI14 %A Patrick Saint-Dizler %T An Approach to Natural-Language Semantics in Logic Programming %J MAG135 %P 329-356 %K AI02 AI10 %A P. H. Schmitt %T Computational Aspects of Three-Valued Logic %B BOOK82 %P 190-198 %K AI11 O04 %A Yoshohito Toyama %T How to Prove Equivalence of Term Rewriting Systems without Induction %B BOOK82 %P 118-127 %K AI11 AI14 %A Jonathan Traugott %T Nested Resolution %B BOOK82 %P 394-402 %K AI11 %A Kyastutis Urba %T Redundancy of Features in a Classification Problem %J Statist. Problemy Upravleniya No. 72 %D 1986 %P 56-63 %K O04 %X Russian with English and Lithuanian Summaries %A Christoph Walther %T A Classification of Many-Sorted Unification Problems %B BOOK82 %P 525-537 %K AI11 AI14 %A Tie Cheng Wang %T ECR: An Equality Conditional Resolution Proof Procedure %B BOOK82 %P 254-271 %K AI11 %A Yuan Yuan Wang %T A Generalized Paramodulation-Resolution Method %J Nanjing Daxue Xuebao Ziran Kexue Ban %V 22 %D 1986 %N 2 %P 205-210 %K AI11 %X Chinese with English Summary %A Richard Cole %A Chee K. Yap %T Shape From Probing %J J. Algorithms %V 8 %D 1987 %N 1 %P 19-38 %K AI06 AI07 %A Peter Hall %A D. M. Titterington %T On Some Smoothing Techniques Used in Image Restoration %J J. Roy. Satist. Soc. Ser. B. %V 48 %D 1986 %N 3 %P 330-343 %K AI06 %A R. Schott %T Nonlinear Filtering and Stochastic Textures %J Math. Modelling %V 8 %D 1987 %P 167-169 %K AI06 %A Miguel Filgueiras %T Cooperating Rewrite Processes for Natural-Language Analysis %J MAG135 %P 299-328 %K AI11 AI02 %A Horst Reichel %T Behavioral Program Specification %B BOOK83 %P 390-411 %K AA08 %A Eugenio Moggi %T Categories of Partial Morphisms and the $lambda sub p$ - Calculus (extended abstract) %B BOOK84 %P 242-251 %K AA08 %A P. Hajek %T Some Conservativeness Results for Nonstandard Dynamic Logic %B BOOK83 %P 443-449 %K AI10 %A Thomas M. Fischer %T On the Average Complexity of Searching for Partial Match Queries in Multidimensional Search Trees %B BOOK83 %P 379-390 %K O06 %A Werner Alexi %T Extraction and Verification of Programs through the Analysis of Formal Proofs %B BOOK81 %P 135-152 %K AA08 %A P. Borowik %A W. Korczynski %A T. Kudla %T An Axiomatic Characterisation of an Algebra of Processes %B BOOK83 %P 141-150 %K AA08 ------------------------------ End of AIList Digest ******************** 14-Jul-87 23:17:59-PDT,21809;000000000000 Mail-From: LAWS created at 14-Jul-87 23:00:54 Date: Tue 14 Jul 1987 22:58-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #181 To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 15 Jul 1987 Volume 5 : Issue 181 Today's Topics: Classification - Natural kinds & Fuzzy Categories, Comment - Need for Harnad-Style Discussions ---------------------------------------------------------------------- Date: 10 Jul 87 1019 PDT From: John McCarthy Subject: Natural kinds Recently philosophers, Hilary Putnam I think, introduced the concept of natural kind which, in my opinion, is one of the few things they have done that is useful for AI. Most nouns designate natural kinds, uncontroversially "bird", and in my opinion, even "chair". (I don't consider "natural kind" to be a linguistic term, because there may be undiscovered natural kinds and never articulated natural kinds). The clearest examples of natural kind are biological species - say penguin. We don't have a definition of penguin; rather we have learned to recognize penguins. Penguins have many properties I don't know about; some unknown even to penguin specialists. However, I can tell penguins from seagulls without a precise definition, because there aren't any intermediates existing in nature. Therefore, the criteria used by people or by the programs we build can be quite rough, and we don't all need to use the same criteria, because we will come out with the same answer in the cases that actually arise. In my view the same is true of chairs. With apologies to Don Norman, I note that my 20 month old son Timothy recognizes chairs and tables. So far as I know, he is always right about the whether the objects in our house are chairs. He also recognizes toy chairs, but just calls them "chair" and similarly treats pictures of chairs in books. He doesn't yet say "real chair", "toy chair" and "picture of a chair", but he doesn't try to sit on pictures of chairs. He is entirely prepared to be corrected about what an object is. For example, he called a tomato "apple" and accepted correction. We should try to make AI systems as good as children in this respect. When a an object is named, the system should generate a gensym, e.g. G00137. To this symbol should be attached the name and what the system is to remember about the instance. (Whether it remembers a prototype or a criterion is independent of this discussion; my prejudice is that it should do both if it can. The utility of prototypes depends on how good we have made it in handling similarities.) The system should presume (defeasibly) that there is more to the concept than it has learned and that some of what it has learned may be wrong. It should also presume (although will usually be built into the design rather than be linguistically represented) that the new concept is a useful way to distinguish features of the world, although some new concepts will turn out to be mere social conventions. Attaching if-and-only-if definitions to concepts will sometimes be possible, and mathematical concepts often are introduced by definitions. However, this is a rare case in common sense experience. I'm not sure that philosophers will agree with treating chairs as natural kinds, because it is easy to invent intermediates between chairs and other furniture. However, I think it is psychologically correct and advantageous for AI, because we and our robots exist in a world in which doubtful cases are rare. The mini-controversy about penguins can be treated from this point of view. That penguins are birds and whales are mammals has been discovered by science. Many of the properties that penguins have in common with other birds have not even been discovered yet, but we are confident that they exist. It is not a matter of definition. He who gets fanatical about arbitrary definitions will make many mistakes - for example, classifying penguins with seals will lead to not finding tasty penguin eggs. ------------------------------ Date: Sat 11 Jul 87 21:45:36-PDT From: Ken Laws Reply-to: AIList-Request@STRIPE.SRI.COM Subject: Natural Kinds I would not be so quick to thank recent philosophers for the concept of natural kinds. While I am not familiar with their contributions, the notion seems similar to "species" in biology and "cluster" in engineering and statistics. Cluster and discriminant analysis go back to at least the 1930s, and have always depended on the tendency of objects under study to group into classes. -- Ken ------------------------------ Date: 13 Jul 87 16:31:17 GMT From: uwslh!lishka@rsch.wisc.edu (Christopher Lishka) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <3930@sunybcs.UUCP> dmark@marvin.UUCP (David M. Mark) writes: >In article <974@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: >> >> >>In Article 185 of comp.cog-eng sher@rochester.arpa (David Sher) of U of >>Rochester, CS Dept, Rochester, NY responded as follows to my claim that >>"Most of our object categories are indeed all-or-none, not graded. A penguin >>is not a bird as a matter of degree. It's a bird, period." -- >> >>> Personally I have trouble imagining how to test such a claim... >> >>Try sampling concrete nouns in a dictionary. > >Well, a dictionary may not always be a good authority for this sort of >thing. I don't want to start a huge discussion on a related topic, but I guess I'll throw in my two-cents worth. Mr. Harnad states that one should try sampling concrete nouns in a dictionary. It seems to me that a short while ago there was some discussion around the country as to what a dictionary's purpose actually is, to which a prominent authority on the subject replied that a dictionary is *only* a description of what people are commonly using certain words for. Now, one upshot of this seems to be that a dictionary, in the end, is NOT a final authority on many words (if not all of them included). It can only provide a current description of what the public in general is using the word for. In the case of some words, many people will use them for many different things. This may be one reason for the problems with the word 'map.' In the case of a penguin, scientifically it is considered a bird. I consider it a bird, although a penguin certainly does not fly in the air. However, if every English-speaking person except a few, say myself and Mr. Harnad, suddenly decided to think of a penguin as something other than a bird, than a dictionary's description would need to be changed, for myself and Mr. Harnad would be far outweighed. I suspect that the dictionary would have some entry as to the historical meaning of 'penguin' (i.e. a penguin used to be considered a bird, but now it is something else). However, since a dictionary is supposed to be descriptive of a language in its current usage, the entry for penguin would have to be modified. Which brings me to my point. Given that a dictionary is a descriptive tool that seeks to give a good view of a language as it is currently being used, can it really be used as a final authority? My feeling is no; just look at all the different uses of a certain word among your friends, not to mention the entire state you live in, not to mention your continent, not to mention the entire English-speaking population of the world. Holy cow! You've suddenly got a lot of little differences in meaning for a certain word. Not to mention slang and local terms (e.g. has anyone ever heard of the word 'bubbler?' It means a 'Water Fountain' here in Wisconsin, but you'd be surprised how many people don't know this term). In this case you can only look at words as a 'graded' term, not an all-or-none term if you are using a dictionary as the basis for a definition. Sure, if you want to use a scientific definition for penguin, go ahead...since science seems to seek to be unambiguos (unlike general spoken language), then you will have a better all-or-none description. But I don't think you can go about using a dictionary, which is a descriptive tool, as an all-or-none decisive authority on what a word means. If I remember back to a Linguistics course I took, this is the same difference as denotation vs. connotation. A couple notes: if you notice above (and right here), I use the word 'you' (as a technical writer would use the word 'one') to refer to a person in general (i.e. the reader). This is not generally accepted as proper English by the people who seek to define proper English, but it is the term that is used by most people that I have known (here in Wisconsin). It seems to me that this is further evidence of my argument above, because I do not think twice in using this term 'you;' it is how I was raised. Also, please don't start a discussion on language in this group unless it pertains to A.I. (and in some case it does); I just felt that someone ought to speak up on the ambiguity of words, and how to different people there might be problems with using a dictionary as a basis for judgement. If you want to continue this discussion, please e-mail me, and I will respond in a decent amount of time (after I cool off in the case of flames ;-) -- Chris Lishka /lishka@uwslh.uucp Wisconsin State Lab of Hygiene <-lishka%uwslh.uucp@rsch.wisc.edu \{seismo, harvard,topaz,...}!uwvax!uwslh!lishka ------------------------------ Date: 10 Jul 87 16:47:54 GMT From: trwrb!aero!venera.isi.edu!smoliar@ucbvax.Berkeley.EDU (Stephen Smoliar) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <3930@sunybcs.UUCP> dmark@marvin.UUCP (David M. Mark) writes: > we conducted >a number of experiments and found many ambiguous stimuli near the boundary >of the concept "map". Air photos and satellite images are an excellent >example: they fit the dictionary definition, and some people feel very >strongly that they *are* maps, others sharply reject that claim, etc. >Museum floor plans, topographic cross-profiles, digital cartographic >data files on tape, verbal driving directions for navigation, etc., are >just some examples of the ambiguous ("fuzzy"?) boundary of the concept >to which the English word "map" correctly applies. I strongly suspect >that "map" is not unique in this regard! Indeed, it almost seems as if "What is a map?" is not really the appropriate question. The better question might be "What can be used as a map?" or perhaps "How can I use a FOO as a map?" Furthermore, I agree that "map" is probably not unique. There are probably any number of bindings for BAR for which "What is a BAR?" runs into similar difficulty and for which "How can I use a FOO as a BAR?" is the more useful question. One candidate I might propose to discuss along these lines is the concept of "algorithm." There are any number of entities which might be regarded as being used as algorithms, ranging from Julia Child's recipies to chromosomes. It would seem that any desire to classify such entities as algorithms is only valuable to the extent that we are interested in the algorithmic properties such entities possess. For example, we might be interested in the nature of recipes which incorporate "while loops" because we are concerned with how such loops terminate. In an earlier posting, Harnad gave the example of how we classify works of art according to particular styles. Such classifications may also be susceptible to this intermediate level of interpretation. Thus, you may or may not choose to view a particular tapestry as an allegory. You may or may not choose to view it as a pastoral. Such decisions influence the way you see it and "parse" it as part of your artistic appreciation, regardless of whether or not your particular view coincides with that of the creator! I suspect there is a considerable amount of such relativity in the way we detect categories. That relativity is guided not by what the categories are or what their features are but by how we intend to put those categories to use. (In other words, the issue isn't "What features are present?" but "What features do we want to be present?") ------------------------------ Date: 14 Jul 87 15:37:00 GMT From: apollo!laporta@beaver.cs.washington.edu (John X. Laporta) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <245@uwslh.UUCP> lishka@uwslh.UUCP (Christopher Lishka) writes: >Given that a dictionary is a descriptive >tool that seeks to give a good view of a language as it is currently being >used, can it really be used as a final authority? My feeling is no; SUMMARY (1) You are absolutely right. There is no 'final authority' because language changes even as one tries to pin it down, with a dictionary, for example. (2) AI programs designed to 'understand' natural language must include an encyclopedic as well as a lexicological (dictionary) competence. (3) The nonexistence to date of perfect artificial understanders of natural language should not be surprising, given the enormity of the task of constructing an artificial encyclopedic competence. (4) The encyclopedia in this instance must grow with the language, preserving past states, simulating present states, and predicting future states. ELABORATION Tackling (2) first: While dictionary definitions are helpful guides in some respects, the nature of linguistic competence is encyclopedic rather than lexicological. For instance, you might hear someone say: Because I was going to give a cocktail party, I went to the mall to buy whiskey, peanuts, and motor oil. A lexicological competence would deem this sentence grammatical and unremarkably consistent, since 'mall' includes the availability of all the items mentioned. An encyclopedic competence, on the other hand, would mark this sentence as strange, since 'motor oil' is not a part of 'cocktail party,' unless, I suppose, you were willing to assume that some of the guests needed mechanical, not social, lubrication. Even this conjecture is unlikely, however, because 'cocktail party' includes humans consuming alcoholic beverages. A case of Billy Beer at the local Exxon is not a cocktail party. Car mechanics do not come to work in little black dresses. An encyclopedic competence is able (a) to isolate the assumptions an utterance requires for coherence, (b) to rank their probability, and (c) thus to evaluate the coherence of the utterance as a whole. Further, 'encyclopedic' in this context includes more than is found in the _Brittanica_. A humorist might write (in the character of a droll garage mechanic) about a parley to negotiate sale of a gas station. He decides to provide a little festive atmosphere by bringing along some beer. But even this hypothesis doesn't eliminate all strangeness: why is the mechanic buying motor oil at the supermarket? Certainly he could get a better price from his distributor. This sentence is a mine of linguistics lessons, but the above should be enough to suggest my point. Encyclopedic competence, however provided, (scripts or semantically marked graphs of words, to give two examples which are not mutually exclusive) is crucial to understanding even the topic of an utterance. The wider question evolves from (1) ... : Language is an elaboration of symbols which refer to other symbols. The 'last stop' (the boundary of semiotic analysis, not the the boundary of the linguistic process itself in actual beings or machines) is the connection of certain signs to 'cultural units.' These pieces of memory are what ground symbol nets to whatever they are grounded upon. (I prefer Harnad's formulation, but that is not crucial for this discussion.) When Og the Caveman remembers one morning the shape of the stone that he used as a scraper yesterday, a cultural unit exists, and stones of that shape are the first signs dependent upon it. To oversimplify, the process continues infinitely as signs are connected to other signs, new cultural units are formed, signs modify other signs, etc. ... and concludes with (3) and (4): Meaning is 'slippery' because language changes as it is used. A historically amnesiac encyclopedic competence for 1980 would mark as improbable sentences used daily at American slave auctions of the 1840's. SOURCE NOTE: Nearly everything I have said here has been elaborated by Umberto Eco in his book 'A Theory of Semiotics' and subsequent writings. ------------------------------ Date: 14 Jul 87 20:25:01 GMT From: ritcv!cci632!dwp@cs.rochester.edu (Dana Paxson) Subject: Re: Thanks. (was Re: Results of Symbol Grounding Poll) In article <1010@mind.UUCP> ghn@mind.UUCP (Gregory Nelson) writes: >In article <993@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: >>[] >>[make the] Net the reliable and respectable medium of scholarly communication >>that I and (I trust) others are hoping it will evolve into. >> ... >>(4) I continue to be extremely enthusiastic about and committed to >>developing the remarkable potential of electronic networks for scholarly >>communication and the evolution of ideas. I take the present votes to >>indicate that the current Usenet Newsgroups may not be the place to attempt >>to start this. > > ... Perhaps you should take some time off to >look at some of the other newsgroups. The comp.xxxx discussions are naturally >oriented to computer people, but things like rec.xxx and sci.xxx are much >more "broadminded" (if you will.) If you want a real surprise, try tuning >in to the Deja Vu discussion on misc.psi or something like that. > I realize that this is belated input. As one who followed along with an occasional understanding of the discussion on symbol grounding, I have been attracted both to the discussion and to the way in which Stevan Harnad conducted it. I admire the discipline and rigor evident in his postings, and see his work as an example of how a newsgroup functioning often as a bulletin board with limited scope can be enriched by some really difficult exploration. Some of the other contributors to the discussion appeared to work well at a level near Mr. Harnad's. It has been an exciting series of exchanges. I regret the loss of the discussion from the newsgroup. Any reader of the most potent material on computer science will find that the authors reach out to many fields to gain inspiration, illustration, and, yes, even forms of grounding(!) for their work. Especially grounding. Like any other science area with meaning, computer science does not begin in words (or bytes) and end in bytes (or words). It ends in application, or at least applicability, to our lives. In the AI realm, that applicability is becoming an intimate metamorphism, a mapping/transformation, of how we work rather than a translation of what we do. If I can characterize an aspect of the symbol grounding discussion, it is a knife-sharp exploration of the type of problem dismissed by so many as having a self-evident solution. This class of problem is precisely the type which is most difficult even to see, let alone solve. Witness the depth and detail of the exchanges we have seen. If others become impatient with the material, they don't have to read it; but this topic area appears to be poorly understood by anybody, and desperately needs close dialogue. Personally, I feel strongly the need to extend my cognitive framework with such powerful and challenging material. Perhaps the outcomes from discussions like this one have too much potential for making a lot of funded thesis work and product development irrelevant... but then some outcomes can unfold whole new realms of exploration and advancement. Unless I am mistaken, these newsgroups can play an active role in this unfoldment. I don't want to see anything this good be relegated to an obscure electronic cranny, or lumped with a lot of diffuse and irrelevant outpourings. Computer scientists have a lot to learn from the symbol-grounding exchanges right here. I sense that there are many quiet readers out there who have powerful ideas relating to this subject, but who have kept silent on seeing contemptuous and abusive complaints of others about the length and content of the postings. For complaints, it seems reasonable to address the complaints to authors privately, or to the moderator if there is one; but open criticism on the net discourages its use by those whose insight and sensitivity exceed their boldness. Making one's views public is an intimidating process in itself, so why should we raise the level of intimidation? For my part, I would like to ask for a citation for Mr. Harnad's original article on the subject of symbol grounding; I want to read it to find out what started the interchange I have seen. I tuned in late in the process. Thanks to all of the participants in this probing discussion. The views expressed here are my own. Dana Paxson Systems Engineering Computer Consoles, Incorporated Rochester, New York 716 482-5000 CIS User ID: 76327,65 ------------------------------ End of AIList Digest ******************** 14-Jul-87 23:27:38-PDT,18937;000000000000 Mail-From: LAWS created at 14-Jul-87 23:23:29 Date: Tue 14 Jul 1987 23:20-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #182 To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 15 Jul 1987 Volume 5 : Issue 182 Today's Topics: Logic Programming - ICOT Prolog Progress, Humor - AI Justification of Star Wars, Speculation - Moravec on Immortality, Philosophy of Science - AI as a Science ---------------------------------------------------------------------- Date: Wed, 15 Jul 87 10:32:20 JST From: Chikayama Takashi Reply-to: chik@icot.icot.JUNET (Chikayama Takashi) Subject: Re: Say, what ever happened to ... ICOT Prolog????? In article <8706111231.AA18169@mitre.arpa> elsaesser%mwcamis@MITRE.ARPA writes: >It seems ages ago that the 5th generation project was going to >reinvent AI in a Prolog "engine" that was to do 10 gazillion " >LIPS". Anyone know what happened? I mean, if you can make so many >"quality" cars (sans auto transmission, useful A/C, paint that can take >rain and sun, etc.), why can't you make a computer that runs an NP-complete >applications language in real time??? Semi-seriously, what is the status >of the 5th generation project, anyone got an update? Well, we are sorry not distributing enough information to the AI society. Most papers related to ICOT's research are distributed to the logic programming society but not to the AI world (I guess you know how poor propagandist Japanese are:-). Many are reported in: International Conference on Logic Programming IEEE Symposium on Logic Programming Please look into proceedings of these conferences. For about 10 gazillion LIPS computers: What our research of these 5 years revealed is that highly parallel hardware can never be practical without much software effort, including new concepts in programming languages. More stress is put upon software than in the original project plan. Indeed, VLSI technology is dropped off from the project. Our experience shows that VLSI technology is NOT the most difficult point in the way to realistic highly parallel computer systems. An efficient system with 256 processors may be built without changing the software at all. But for systems with 4096 processors, we need a drastic change. And this is what we need to achieve 10 gazillion LIPS. NOT that VLSI technology has become easier, but that we have found MORE difficult problems, unfortunately. Where are we? Well, one of our recent hardware achievement is the development of the PSI-II machine, which executes 400 KLIPS (much less than 10 gazillion, I guess :-). It is a sequential machine and will be used as element processors of our prototype parallel processor Multi-PSI V2 (with 64 PE's), whose hardware is scheduled to come up at the end of this year. If you are interested in our research, a survey by myself titled: "Parallel Inference System Researches in the FGCS Project" will be presented in the IEEE Symposium on Logic Programming, held at San Francisco during Aug 31-Sep 4, 1987. If you are more interested in our project, please join the FGCS'88 conference. It will be held in Tokyo during Nov 28-Dec 2, 1988. Takashi Chikayama ------------------------------ Date: 14-Jul-1987 2028 From: minow%thundr.DEC@decwrl.dec.com (Martin Minow THUNDR::MINOW ML3-5/U26 223-9922) Subject: Book Report From "Dirk Gently's Holistic Detective Agency," by Douglas Adams. (New York: Simon and Schuster, 1987): "Well," he said, "it's to do with the project which first made the software incarnation of the company profitable. It was called _Reason_, and in its own way it was sensational." "What was it?" "Well, it was a kind of back-to-front program. It's funny how many of the best ideas are just an old idea back-to-front. You see, there have already been several programs written that help you make decisions by properly ordering and analysing all the relevant facts.... The drawback with these is that the decision which all the properly ordered and analyzed facts point to is not necessarily the one you want. "... Gordon's great insight was to design a program which allowed you to specify in advance what decision you wished it to reach, and only then to give it all the facts. The program's task, ... was simply to construct a plausible series of logical-sounding steps to connect the premises with the conclusion." .... "Heavens. and did the program sell very well?" "No, we never sold a single copy.... The entire project was bought up, lock, stock, and barrel, by the Pentagon. The deal put WayForward on a very sound financial foundation. Its moral foundation, on the other hand, is not something I would want to trust my weight to. I've recently been analyzing a lot of the arguments put forward in favor of the Star Wars project, and if you know what you're looking for, the pattern of the algorithms is very clear. "So much so, in fact, that looking at Pentagon policies over the last couple of years I think I can be fairly sure that the US Navy is using version 2.00 of the program, while the Air Force for some reason only has the beta-test version of 1.5. Odd, that." ------------------------------ Date: Wed 8 Jul 87 16:19:25-PDT From: Ken Laws Subject: Moravec on Immortality [Forwarded with permission of Hans.Moravec@ROVER.RI.CMU.EDU.] From AP Newsfeatures, June 14, 1987 By MICHAEL HIRSH Associated Press Writer PITTSBURGH (AP) - If you can survive beyond the next 50 years or so, you may not have to die at all - at least, not entirely. [...] Hans Moravec, director of Mobile Robot Laboratory of the Robotics Institute at Carnegie Mellon University, believes that computer technology is advancing so swiftly there is little we can do to avoid a future world run by superintelligent robots. Unless, he says, we become them ourselves. In an astonishingly short amount of time, scientists will be able to transfer the contents of a person's mind into a powerful computer, and in the process, make him - or at least his living essence - virtually immortal, Moravec claims. ''The things we are building are our children, the next generations,'' the burly, 39-year-old scientist says. ''They're carrying on all our abilities, only they're doing it better. If you look at it that way, it's not so devastating.'' [...] ''I have found in traveling throughout all of the major robotics and artificial intelligence centers in the U.S. and Japan that the ideas of Hans Moravec are taken seriously,'' says Grant Fjermedal, author of ''The Tomorrow Makers,'' a recent book about the future of computers and robotics. [He] Devotes the first five chapters of his book to the work of Moravec and his proteges at CMU. MIT's Gerald J. Sussman, who wrote the authoritative textbook on artificial intelligence, agreed that computerized immortality for people ''isn't very long from now.'' ''A machine can last forever, and even if it doesn't you can always make backups,'' Sussman told Fjermedal. ''I'm afraid, unfortunately, that I'm the last generation to die. Some of my students may manage to survive a little longer.'' [...] CMU's Alan Newell, one of the so-called founding fathers of artificial intelligence, cautions that while little stands in the way of intelligent machines, the transfer of a human mind into one is ''going down a whole other path.'' ''The ability to create intelligent systems is not at all the same as saying I can take an existing mind and capture what's in that mind. You might be able to create intelligence but not (capture) the set of biological circumstances that went into making a particular mind,'' he says. In Moravec's forthcoming book, ''Mind Children,'' he argues that economic competition for faster and better information-processing systems is forcing the human race to engineer its own technological Armageddon, one that a nuclear catastrophe can only delay. Natural evolution is finished, he says. The human race is no longer procreating, but designing, its successors. ''We owe our existence to organic evolution. But we owe it little loyalty,'' Moravec writes. ''We are on a threshold of a change in the universe comparable to the transition from non-life to life.'' Moravec's projections are based on his research showing that, on the average, the cost of computation has halved every two years from the time of the primitive adding machines of the late 19th century to the supercomputers of the 1980s. [...] Moreover, the rate is speeding up, and the technological pipeline is full of new developments, like molecule-sized computer circuits and recent advances in superconductors, that can ''sustain the pace for the foreseeable future,'' he says. The implications of a continued steady decrease in computing costs are even more mind-boggling. It is no surprise that studies in artificial intelligence have shown sparse results in the last 20 years, Moravec says. Scientists are severely limited by the calculating speed and capacity of laboratory computers. Today's supercomputers, running at full tilt, can match in power only the 1-gram brain of a mouse, he says. But by the year 2010, assuming the growth rate of the last 80 years continues, the best machines will be a thousand times faster than they are today and equivalent in speed and capacity to the human mind, Moravec argues. [...] ''All of our culture can be taken over by robots. It'll be boring to be human. If you can get human equivalence by 2030, what will you have by 2040?'' Moravec asks, laughing. ''Suppose you're sitting next to your best friend and you're 10 times smarter than he is. Are you going to ask his advice? In an economic competition, if you make worse decisions, you don't do as well,'' he says. ''We can't beat the computers. So that opens up another possibility. We can survive by moving over into their form.'' There are a number of different scenarios of ''digitizing'' the contents of the human mind into a computer, all of which will be made plausible in the next 50 to 100 years by the pace of current technology, Moravec says. One is to hook up a superpowerful computer to the corpus callosum, the bundle of nerve fibers that connects the two hemispheres of the brain. The computer can be programmed to monitor the traffic between the two and, eventually, to teach itself to think like the brain. After a while, the machine begins to insert its own messages into the thought stream. ''The computer's coming up with brilliant solutions and they're just popping into your head,'' Moravec says [...] As you lose your natural brain capacity through aging, the computer takes over function by function. And with advances in brain scanning, you might not need any ''messy surgery,'' Moravec says. ''Perhaps you just wear some kind of helmet or headband.'' At the same time, the person's aging, decrepit body is replaced with robot parts. ''In the long run, there won't be anything left of the original. The person never noticed - his train of thought was never interrupted,'' he says. This scenario is probably more than 50 years away, Moravec says, but because breakthroughs in medicine and biotechnology are likely to extend people's life spans, ''anybody now living has a ticket.'' Like many leading artificial intelligence researchers, Moravec discounts the mind-body problem that has dogged philosophers for centuries: whether a person's identity - in religious terms, his soul - can exist independently of the physical brain. ''If you can make a machine that contains the contents of your mind, then that machine is you,'' says MIT's Sussman. Moravec believes a machine-run world is inevitable ''because we exist in a competing economy, because each increment in technology provides an advantage for the possessor . . . Even if you can keep them (the machines) slaves for a long time, more and more decision-making will be passed over to them because of the competitiveness. ''We may be still be left around, like the birds. It may well be that we can arrange things so the machines leave us alone. But sooner or later they'll accidently step on us. They'll need the material of the earth.'' Such talk is dismissed as sheer speculation by Moravec's detractors, among them his former teacher, Stanford's John McCarthy, who is also one of the founding fathers of artificial intelligence research. McCarthy says that while he respects Moravec's pioneering work on robots, his former Ph.D student is considered a ''radical.'' ''I'm more uncertain as to how long it (human equivalence) will take. Maybe it's five years. Maybe it's 500. He has a slight tendency to believe it will happen as soon as computers are powerful enough. They may be powerful enough already. Maybe we're not smart enough to program them.'' Even with superintelligent machines, McCarthy says, it's hardly inevitable that computers will take over the world. ''I think we ought to work it out to suit ourselves. In particular it is not going to be to our advantage to give things with human-level intelligence human-like emotions (like ambition). You might want something to sit there and maybe read an encyclopedia until you're ready to use it again,'' he says. George Williams, an emeritus professor of divinity at Harvard University, called Moravec's scenario ''entirely repugnant.'' [...] McCarthy, however, insists there's no need to panic. ''Because the nature of the path that artificial intelligence will take is so unknown, it's silly to attempt to plan any kind of social policy at this early time,'' he says. ------------------------------ Date: Sun 12 Jul 87 19:45:34-PDT From: Lee Altenberg Subject: AI is not a science This discussion has brought to my mind the question of undecidability in cellular automata, as discussed by S. Wolfram. For some rules and initial sequences , the most efficient way of finding out how the automaton will behave is simply to run it. Now, what is the status of knowledge about the behavior of automata and the process of obtaining this knowledge? Is it a science or not? Invoking some of the previous arguments regarding AI, it could be said that it is not a science because knowing something about an automaton tells one nothing about the actual world. That is why mathematics has been called not science. Yet, to find out how undecidable automata behave one needs to carry out experiments of running them. In this way they are just like a worldly phenomenon where knowledge about them comes from observing them. One must take an empirical approach to undecidable systems. But there is another angle of evaluation. Naturalists have been belittled as "not doing science" because their work is largely descriptive. Does science consist then in making general statements? Or to be more precise, does science consist of redescribing reality in terms of some general statements plus smaller sets of statements about the world, which when combined can generate the full (the naturalists's) description of reality? If this is to be the case, then all examples of undecidable (and chaotic, I would guess) processes fall outside the dominion of science, which seems to me overly restrictive. ------------------------------ Date: Mon, 13 Jul 87 15:08:07 bst From: Stefek Zaba Subject: AI as science: establishing generality of algorithms In response to the points of Jim Hendler and John Nagle, about whether you can verify that your favourite planning system can be shown to be more general than the Standard Reference: At the risk of drawing the slings and arrows of people who sincerely believe Formalism to be the kiss of death to AI, I'd argue that there *are* better characterisations of the power of algorithms than a battery of test cases - or, in the case of the typical reported AI program, described in necessarily space-limited journals, a tiny number thereof. Such characterisations are in the form of more formal specs of the algorithm - descriptions of it which strip away implementation efficiency tricks, and typically use quantification and set operations to get at the gist of the algorithm. You can then *prove* the correctness of your algorithm *under given assumptions*, or "equivalently" derive conditions under which your algorithm produces correct results. Such proofs are usually (and, I believe, more usefully) "rigorous - but - informal"; that is a series of arguments with which your colleagues cannot find fault, rather than an immensely long and tortuous series of syntactic micro-steps which end up with a symbol string representing the desired condition. Often it's easier to give sufficient (i.e. stronger than necessary) conditions under which the algorithm works than a precise set of necessary-and-sufficient ones. *Always* it's harder (for mere mortals like me, anyway) than just producing code which works on some examples. An example of just such a judicuious use of formalism which I personally found inspiring is Tom Mitchell's PhD thesis covering the version space algorithm (Stanford 1978, allegedly available as STAN-CS-78-711). After presenting a discursive description of the technique in chapter 2, chapter 3 gives a formal treatment which introduces a minimum of new terminology, and gives a simple and *testable* condition under which the algorithm works: "A set of patterns P with associated matching predicate M is said to be an admissible pattern language if and only if every chain [totally ordered subset] of P has a maximum and a minimum element". Stefek Zaba, Hewlett-Packard Labs, Bristol, England. [Standard disclaimer concerning personal nature of views applies] ------------------------------ Date: 13 Jul 87 05:23:56 GMT From: ihnp4!lll-lcc!esl.ESL.COM!ssh@ucbvax.Berkeley.EDU (Sam) Reply-to: ssh@esl.UUCP (Sam) Subject: is AI a science? [There are several components of AI, as there are of CS, but...] Let's take a step back. Is "Computer Science" a science? -- Sam ------------------------------ End of AIList Digest ******************** 16-Jul-87 23:19:35-PDT,20285;000000000001 Mail-From: LAWS created at 16-Jul-87 23:08:59 Date: Thu 16 Jul 1987 23:05-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #183 To: AIList@STRIPE.SRI.COM AIList Digest Friday, 17 Jul 1987 Volume 5 : Issue 183 Today's Topics: Philosophy - Natural Kinds & Philosophy of Science & Categorization & Symbol Grounding ---------------------------------------------------------------------- Date: 15 Jul 87 14:19 PDT From: Tony Wilkie /DAC/ Subject: Natural Kinds I may get sizzled for this, but I will suggest that the term "natural kind", while a fairly recent addition to the philosophical lexicon, is a conceptual descendant of Plato`s Forms, and more closely approximated in meaning to Aristotle's discussions of 'kinds' in his Metaphysics. Chairs would certainly be a paradigm example of a Platonic Form, and Aristotle in his Metaphysics used his horse, Bucephalus, as an example in his discussion of kinds. Given his inclination as sort of a teleological guerilla, Aristotle would have (and may have) had a tough time separating his 'kinds' concept from 'species' in the biological cases. Still, I think it safe to say that philosophical discussion of ontology preceded the development of a formal concept of species. Tony L. Wilkie ------------------------------ Date: Thu, 16 Jul 87 15:42:11 EDT From: mclean@nrl-css.arpa (John McLean) Subject: Natural Kinds Even "recent" philosophical discussions of natural kinds go back 20 years and much further if you count Nelson Goodman's stuff on projectibility of predicates (why do we assume emeralds are green and not grue, i. e., green until the year 2000 and then blue?) or much of the stuff written in response to Hempel's problem whether a nonblack nonraven could could count as a confirming instance of the claim that all ravens are black (since the claim that all P's are Q's is logically equivalent to the claim that all nonQ's are nonP's). But I think you can also view much of what Plato had to say about forms and what Aristotle had to say about substance as being concerned with the problem of natural kinds as well. However, I think the issue being raised about recognizing penguins, chairs, etc. goes back to Wittgenstein's _Philosophical_Investigations_: For if you look at them you will not see something that is common to all, but similarities, relationships, and whole series of them at that...I can think of no better expression to characterize these similarities than "family resemblance"... John McLean ------------------------------ Date: 16 Jul 87 2207 PDT From: John McCarthy Subject: re: AIList Digest V5 #181 [In reply to message sent Tue 14 Jul 1987 22:58-PDT.] The distinction I had in mind between natural kind and cluster is the presumed existence of as yet unknown properties of a natural kind. When I said "doubtful cases are rare", I left myself open to misunderstanding. I meant that in case of chairs in Timothy's experience doubtful cases are rare. Therefore, for a child to presume a natural kind on hearing a word or seeing an object is advantageous, and it will also be advantageous to built AI systems with this presumption. Finally, a remark concerning the "symbol grounding" discussion. My problems with it were mainly quantitative - there was just too much to follow. I suspect that Stevan Harnad's capacity to follow very long discussions is exceptional. I would welcome a summary of the different points of view by someone who did follow it and feels himself sufficiently uncommitted to any single point of view. ------------------------------ Date: Thu, 16 Jul 87 17:18 EDT From: Nichael Cramer Reply-to: Nichael Cramer Subject: AIList Digest V5 #182 >> >> Let's take a step back. Is "Computer Science" a science? -- Sam >> There is the old chestnut that one should be leery of any disipline that feels such a need to justify itself that it appends the term "Science" to its own name. Witness "Social Science". Or more to the point, "Creation Science" [sic]. [Standard disclaimer concerning personal nature of views applies] NICHAEL Rednecks for Rainforest ------------------------------ Date: 14 Jul 87 22:20:56 GMT From: mcvax!botter!klipper!biep@seismo.css.gov (J. A. "Biep" Durieux) Subject: Definition of science and of scientific method. 1) I think this discussion belongs in sci.philosophy.tech, and perhaps in sci.research, but definitely not in any of the other groups. Please let's move out of the wrong newsgroups. This article is meant as a merger of two discussions, one in sci.med (and other places), and one in comp.ai. Followups will go to sci.philosophy.tech *only*. 2) There are multitudes of definitions for science, and even more usages. Here I talk just about a rather generally accepted stance. 3) There is craft (what engineers and the like do), art (about which I don't want to speak), science (the methodically unraveling of the secrets of the world ("world" in a broad sense), and philosophy (the necessary building of footholds, standing on which science can be done). 4) Philosophy starts with quarreling about whether God exists, then whether I exist (some say the other way round - for "God" some read "anything at all"), then whether an outside world exist, then how we should look at that world (yielding things like epistemology, ethics, aesthetics, etc.), and, choosing epistemology, which ways of getting knowledge are there and which ones have which value. One of these methods (as many philosophers hold) is reason, and there come logic and mathematics around the corner. Still much dispute (intuitionism for example - could you give us an intro, Lambert Meertens? - or "what constitutes a proof", "what is `mathematical rigour'", etc.) and uncertainty (liars paradox) around, as the means of thinking are still being defined, so they cannot be used freely yet. Perhaps that is a good working definition of science: thinking there where the means for thinking are not yet finished. 5) Science starts (or: sciences start) from the results of the philosophers' work (unhappily the philosophers aren't ready yet, so those results are not as sure as they should be, and certainly not as sure as they are often thought to be by non-philosophical scientists) exploring the world. 6) The definition of "science", and of scientific method, is by its very nature a philosophical, not a scientifical matter. Otherwise one would get paradoxes like: Ockhams razor tells us to throw away any non-necessary principles. The principle of Ockhams razor is non-necessary. So let's throw away Ockhams razor. (Happily, the director of the British Museum will not let you touch it, but anyway, the case is clear.) 7) The above is highly simplified, but I believe that simple introductions are wanting on usenet. Too often I fall into a discussion which supposes knowledge I don't have, of I see some participants don't have. 8) If this spawns serious discussion (only in sci.philosophy.tech, please!) I would be more than pleased. -- Biep. (biep@cs.vu.nl via mcvax) Unix is a philosophy, not an operating system. Especially the latter. ------------------------------ Date: 15 Jul 87 15:45:00 GMT From: apollo!laporta@beaver.cs.washington.edu (John X. Laporta) Subject: Re: The symbol grounding problem: "Fuzzy" categories? In article <3183@venera.isi.edu> smoliar@vaxa.isi.edu.UUCP (Stephen Smoliar) writes: >There are probably any number of bindings for BAR >for which "What is a BAR?" runs >into ... difficulty and for which >"How can I use a FOO as a BAR?" is the more useful >question. > >In >an earlier posting, Harnad gave the example of how we classify works >of >art ... Such classifications may also be susceptible to this >intermediate level of >interpretation. Thus, you may or may not >choose to view a particular tapestry >as an allegory ... [or] as a >pastoral. Such decisions influence the way you see it and >"parse" it >as part of your artistic appreciation, regardless of whether or not >your >particular view coincides with that of the creator! > >I suspect there is a considerable amount of such relativity in the way we >detect categories. That relativity is guided not by what the categories >are or what their features are but by how we intend to put those >categories to use. (In other words, the issue isn't "What features >are present?" but "What features do we want to be present?") Umberto Eco writes in "Euge`ne Sue and _Les Myste`res de Paris_" about this problem. Sue was a sort of gentleman pornographer in post-Napoleonic France. One of his series, about a character like the Shadow who worked revenge on decadent aristocratic evildoers, with a lot of bodice-ripping along the way, caught on with the newly literate general working public. They consumed his book in vast quantities and took it as a call to arms so seriously that Paris was barricaded by people inspired by it. A sex-and-violence pornographic thriller became a call to political reform and the return of morality. The relevant semiotic category is "closure." Roughly speaking, a closed work is one that uses a tight code to tell a tale to an audience sharply defined by their sharing of that code. Superman Comics is an example of a closed work. (There is an entertaining study somewhere of explanations offered by New Guinean tribesmen of a Superman Comic.) Closed works don't ring, so to speak, with the resonance of the entire semiotic continuum, while open works do. Closed works are thus easily subject to gross misinterpretation by readers who don't share the code in which those works are written. Open works, on the other hand, enforce their own interpretation. While there is drift over time in these interpretations, it is far smaller than the vastly divergent interpretations offered of closed works by varying interpreters in the same era. Open works connect to the entire semiotic continuum - indeed, the (broadly) rhetorical methods (tropoi) they use bespeak a purpose of educating the reader about the subjects (topoi) they treat. _Remembrance of Things Past_ is an example of an open work. While a great deal of unfamiliar material and controversial analysis is offered to any reader of those 3000 pages, the mere act of reading them enforces what is, for the purpose of semiotics, a uniform interpretation (read disambiguated topical hypothesis). It is very easy to 'use' a closed work by correlating the elements of an external symbol system with the opaque code the work presents. Of course, if the 'grounding' of one's symbol system bears no relation to that which the work employs, one is just as much 'used' by the work as a consequence. (Imagine, for example, using a rectangular bar of plastic explosive as a straightedge.) It is far more difficult to impose an arbitrary interpretation on an open work, since it contains material that tends to contradict incorrect or incomplete hypotheses about its topos. For example, while we are 'told' that Superman comes from the planet Krypton, etc., we learn by watching Marcel what his origins are, and while Superman comes as a given from space, Marcel's character defines itself in our consciousness by our 'observation' of his life. Furthermore, while Superman is always Superman, Marcel has an origin and a destiny. Marcel changes with time, he breaks with Albertine; Superman always almost, but actually never marries Lois Lane. (Spiderman's recent marriage to Mary Jane is an interesting twist. Certainly by comparison with Superman's, Spiderman's story is an open work.) Historians who based hypotheses about 20th century American atittudes on an analysis of Superman comics would have to confirm them by considerable reference to external sources, while students of early 20th century France would likely use _Remembrance of Things Past_ to confirm their ideas. IN SUMMARY: The relativity of categorization is an inverse index of the 'openness' of the thing categorized. Dr. Morbius in "Forbidden Planet" was able to divine the purpose of Krell instrumentation because the science on which it was founded, while more advanced than his own, shared the same basis in physical reality and hypothesis testing. The space-given monolith in "2001" is indecipherable (a real 'black box', but with undefined input and output), and thus can be 'used' for any purpose at all. ------------------------------ Date: 13 Jul 87 18:16:06 GMT From: linus!philabs!sbcs!bnl!allard@husc6.harvard.edu (rick allard) Subject: Re: The symbol grounding problem: Again... grounding? In article <931@mind.UUCP> harnad@mind.UUCP (Stevan Harnad) writes: >Categorization preformance (with all-or-none categories) is highly reliable >(close to 100%) and MEMBERSHIP is 100%. ... Why add this clause about "real" membership? Isn't the bulk of the discussion about us humble humans doing the categorizing? If we do start wondering about this larger realm, does it bear on categorizing? Rick -- ooooooooooooootter#spoon in bowl !!!!!!!!!!!!& RooM & !!!!!!!!!!!!R oooo M ------------------------------ Date: 15 Jul 87 19:08:35 GMT From: diamond.bbn.com!aweinste@husc6.harvard.edu (Anders Weinstein) Subject: Re: The symbol grounding problem meta-discussion Since I will shortly be posting a follow-up to Harnad's last reply to me on the SGP, I guess I ought to address the meta-discussion. I think that different standards apply in the two domains in which this discussion has been taking place. I recognize that AI-List subscribers rightfully expect some selectivity from an edited digest, and I will understand completely if the moderator chooses not to redistribute my follow-up because the volume on this subject has exceeded the limits demanded by his readership. On the other hand, I see no justification for attempting to squelch the discussion on the Usenet side of things (from which I am participating). This unmoderated forum is avowedly anarchic, and the wishes of a supposed majority are irrelevant -- perhaps no single topic interests a majority of readers. If you're not interested in a discussion that's clearly appropriate for this newsgroup, the right thing to do is just ignore it. The software makes it easy to "kill" a topic you don't care about; do so, and you'll never even *see* the messages. I really don't understand the problem. Anders Weinstein BBN Labs ------------------------------ Date: 15 Jul 87 20:00:29 GMT From: diamond.bbn.com!aweinste@husc6.harvard.edu (Anders Weinstein) Subject: Re: The symbol grounding problem In a previous message, I was prompted by Stevan Harnad's postings to try to explain something I find very interesting, namely, why the psychology of categorical perception won't do much to illuminate the difficult question of how formal symbols should be semantically interpreted, i.e. what the symbols really *mean*. Harnad sent a long reply (message 972@mind.UUCP) explaining the nature of his approach in great detail. The upshot, I think, is that in spite of some of the rhetoric about "symbol grounding", Harnad's project is not really *attempting* to do any such thing. It merely aims to discover the mechanisms underlying certain recognition skills. Since this more modest aim was precisely what I was urging, I am satisfied that there is no major disagreement between us. I want to make clear that I am not here trying to pose any *objection* to Harnad's model considered as a bit of psychology. I am only trying to downplay its significance for philosophical issues. Remember that the traditional conception of "meanings" or "concepts" involves certain properties: for example, meanings are supposed to contain a criterion which determines the correct application of the term, in effect defining the metaphysical essence of the concept in question; they are supposed to serve as elementary constituents of more complex concepts and thoughts; and they are supposed to license analytic implications, such as "all bachelors are unmarried". Since none of these properties seem to be required of the representations in Harnad's theory, it is in a philosophical sense *not* a theory of "concepts" or "meanings" at all. As Harnad should be be happy to concede. But I want to emphasize again an important reason for this which Harnad seemed not to acknowledge. There is a vast difference between the quick, observational categorization that psychologists tend (rightly) to focus on and the processes involved in what might be called "conclusive" classification. This is the difference between the ability to recognize something as fish-like in, say, 500 milliseconds, and the ability to ascertain that something *really* is a fish and not, say, an aquatic mammal. Now the former quick and largely unconscious ability seems at least a plausible candidate for revealing fundamental cognitive mechanisms. The latter, however, may involve the full exercise of high-level cognition -- remember, conclusive classification can require *years* of experiment, discussion and debate, and potentially involves everything we know. The psychology of conclusive categorization does *not* deal with some specialized area of cognition -- it's just the psychology of all of science and human rationality, the cognitive scientist's Theory of Everything. And I don't expect to see such a thing any time soon. Confusion can result from losing sight of the boundary between these two domains, for results from the former do not carry over to the latter. And I think Harnad's model is only reasonably viewed as applying to the first of these. The rub is that it seems that the notion of *meaning* has more to do with what goes on in the second. Indeed, what I find most interesting in all this is the way recent philosophy suggests that concepts or meanings in the traditional sense are essentially *outside* the scope of forseeable psychology. Some other replies to Harnad: Although my discussion was informed by Quine's philosophy in its reference to "meaning holism", it was otherwise not all that Quinean, and I'm not sure that Quine's highly counter-intuitive views could be called "standard." Note also that I was *not* arguing from Quine's thesis of the indeterminacy of translation; nor did I bring up Putnam's Twin-Earth example. (Both of these arguments would be congenial to my points, but I think they're excessively weighty sledgehammers to wield in this context). The distinction between observational and "conclusive" classification, however, does bear in mind Putnam's points about the the non-necessity of stereotypical properties. I also don't think that philosophers have been looking for "the wrong thing in the wrong way." I think they have made a host of genuine discoveries about the nature of meaning -- you cite several in your list of issues you'd prefer to ignore. The only "failure" I mentioned was the inability to come up with necessary and sufficient definitions for almost anything. (Not at all, by the way, a mere failure of "introspection".) I *do* agree that the aims of philosophy are different than those of psychology. Indeed, because of this difference of goals, you shouldn't feel you have to argue *against* Quine or Putnam or even me. You merely have to explain why you are side-stepping those philosophical issues (as I think you have done). And the reason in brief is that philosophers are investigating the notion of meaning and you are not. Anders Weinstein BBN Labs ------------------------------ End of AIList Digest ******************** 19-Jul-87 21:57:19-PDT,17223;000000000001 Mail-From: LAWS created at 19-Jul-87 21:47:43 Date: Sun 19 Jul 1987 21:41-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #184 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 20 Jul 1987 Volume 5 : Issue 184 Today's Topics: Queries - Cooperating Expert Systems & Garbage Collection Suppression, Comments - Expert System for Rocket Launching & Automatic Implementation of Abstract Specifications & ANIMAL in BASIC & Immortality via Computer, Correction - Spang Robinson Report, June 1987, Perception - Natural Kinds ---------------------------------------------------------------------- Date: Wed, 15 Jul 87 01:24 EDT From: Arnold@DOCKMASTER.ARPA Subject: Query - Cooperating Expert Systems I am looking for information on cooperating expert systems. Any pointers, references, etc would be appreciated. Terry S. Arnold Merdan Group 4617 Ruffner St. San Diego CA 92111 Arnold -at Dockmaster (619) 571-8565 ------------------------------ Date: Fri, 17 Jul 87 08:32:41 edt From: nancy@grasp.cis.upenn.edu (Nancy Orlando) Subject: Garbage Collection Suppression Are there any "accepted" methods of writing code that minimize a LISP's tendancy to garbage-collect? I don't mean a switch to turn it off; just a means of minimizing the need for it. I'm dealing particularly with DEC VAX lisp. I have assumed that iteration as opposed to recursion was one way; is this correct? Are there other techniques? Nancy Sliwa nancy@grasp.cis.upenn.edu or nesliwa%telemail@orion.arpa ------------------------------ Date: 15 Jul 87 16:42:47 GMT From: jbn@glacier.STANFORD.EDU (John B. Nagle) Reply-to: jbn@glacier.UUCP (John B. Nagle) Subject: Re: bm654 - Spang Robinson 3#6, 6/87 >Rome Air Force Development Center is building a system to help decide >if foreign rocket launches are threats. I saw the RFP for that one go by when I was at Ford Aerospace. I recommended that we not bid, pointing out that an expert system to make launch-on-warning decisions was a singularly bad idea. Seen in that light, no one at Ford wanted to have anything to do with the program. Nevertheless, RADC apparently found somebody willing to spend their money. Fortunately, most of what RADC funds never gets deployed. John Nagle ------------------------------ Date: 14 Jul 87 16:00:32 GMT From: eagle!icdoc!esh@ucbvax.Berkeley.EDU (Edward Hayes) Subject: Re: Automatic implementation of abstract specifications I just saw an article giving an inexact reference to an MIT technical report by MK Srivas, The exact reference (I just happened to have it on my desk) is: MIT/LCS/TR-276 Automatic Synthesis of Implementations for Abstract Data Types from Algebraic Specifications Mandayam K Srivas June 1982 - hope this is of help. ------------------------------ Date: 17 Jul 87 06:30:19 GMT From: psivax!polyslo!mshapiro@seismo.CSS.GOV (Mitch Shapiro) Reply-to: psivax!polyslo!mshapiro@seismo.CSS.GOV (Mitch Shapiro) Subject: Re: ANIMAL in BASIC ??? In article <8707090304.AA15222@humu.ARPA> dbrauer@humu.UUCP (David L. Brauer) writes: >Somewhere in the darkest reaches of my memory I recall seeing a listing >of the game ANIMAL in BASIC. It's that old standby introduction to rule-based >reasoning that tries to deduce what animal you have in mind by asking >questions like "Does it have feathers?", "Does it have hooves?" etc. There was originally shipped with the Apple II's (maybe for subsequent machines as well) that very program written in BASIC. It learned new animals and stored them in a text file (I think). But it did learn learn them. Find someone you know who has an Apple II. I believe this was shipped with DOS 3.1. -- Yes, I have a pretty old Apple. #7919 just in case anyone out there cares. Mitch Shapiro mshapiro@polyslo (well, for all of another 3 days, that is.) "It has been said that when Science climbs the crest of the hill, it will see that religion has been sitting there all along." --- Dr. Harry Wolper ------------------------------ Date: 15 Jul 87 19:12:42 GMT From: David L. Brauer Reply-to: dbrauer@humu.nosc.mil.UUCP (David L. Brauer) Subject: Re: ANIMAL in BASIC ??? Thanks to all who responded to my request for pointers to Animal in BASIC. The listing can be found in 101 BASIC Games by David H. Ahl. There also may be a version on one of the Apple DOS distributions, although I haven't found it yet. Please, no more lectures on why Animal should not be called a rule-based or expert system. I'm aware that it is a simple tree traversal algorithm. Merely a misnomer on my part. I thought I had seen the listing in an "Intro to AI" slick, that is why I worded the request that way. David C. Brauer MilNet: dbrauer@NOSC.mil ------------------------------ Date: Fri, 17 Jul 87 08:39 EST From: MNORTON%rca.com@RELAY.CS.NET Subject: Re: Immortality via Computer Concerning the AP story on attaining immortality via computers, readers of AIList intrested in thinking more about this may wish to read Fredrick Pohl's new book, "Annals of the Heechee", the forth book in the series which began with "Gateway." Mr. Pohl explores some of the implications of computer subsumption of consciousness, which he calls 'vastening' in the story. Some of the topics touched on include altered preception of reality, differing time-rates between biologicals and computers, and non-corporeal being. Mark J. Norton, RCA Advanced Technology Laboratories, AI Lab. ------------------------------ Date: Wed, 15 Jul 1987 19:31 CST From: Leff (Southern Methodist University) Subject: Corrections Response to errors discovered by Linda Mead: In the summary of the June 1987 Spang Robinson Report, the following corrections should be noted: 1) "The pilot's associate project aims to produce a refrigerator sized computing system, having functionality comparable to a 3-inch by 5-inch checklist card." The d in "card" was missing. 2) Charles Anderson was not precisely identified: He is a Lt. Col., deputy of technology development for SDI in the Command and Control Directorate at Rome Air Development Center at Griffis Air Force Base. 3) The statment regarding "AI research for SDI" was that it "would be relatively nil for awhile." No specific statement on it's "use" was made by Spang Robinson Report. (The paragraph on this subject in the summary had an extraneous double quote character due to a typo. A direct quote was not made. ------------------------------ Date: 17 Jul 87 13:33:46 GMT From: mcvax!botter!hansw@seismo.css.gov (Hans Weigand) Subject: Re: natural kinds It seems to me that _at least_ three kinds of "natural kinds" should be distinguished: (1) genetic kinds, existing by virtue of reproduction ("a horse is a horse because it is born from a horse") Examples: animal and vegetable species (2) mimetic kinds, existing by virtue of imitation, to be subdivided in (a) iconic kinds (by causally determined representation) (the "Xerox-principle" of Dretske: an image of an image of x is again an image of x) Examples: all linguistic symbols (graphic or phonemic) (b) artificial kinds (by imitation on purpose), existing by virtue of preconceived design followed by numerous production (the "Ford-principle" |-) ) Examples: car models, coins (c) fashion kinds (by copying behavior, largely uncontrolled) Examples: social groups (punks, yuppies, ..), styles of art, etc. (3) anthropic/functional kinds, existing by virtue of readiness_to_hand Examples: chair, cup, house, knife, game The last one needs some comments. Each human being needs certain things in order to survive and live in a satisfactory way. These things are mainly determined by the functioning of the human body and community, although there are also environmental and historical-cultural influences. Thus we may recognize an Eskimo iglo, and an African pile-dwelling both as "houses". I think it is not so much the form (iconicity) that matters, but rather that we feel that, when we would live in Greenland (resp. the jungle), we would naturally appreciate or use these things as houses too (to protect us against cold, dangers). Similar arguments can be made for chair etc.. Moreover, (3) combines with (2). We are born into a human society. Our parents had the same needs as we have, so each generation copies these "anthropic kinds" and transfers them to a next generation. This makes it the more easy to recognize a (say Western) house. [In most discussions on "family kinds" and so on, (2) and (3) are not properly distinguished]. "Don't ask what a kind _is_, but rather how it _persists_" Hans Weigand (hansw@cs.vu.nl) ------------------------------ Date: Sat 18 Jul 87 15:17:43-CDT From: Robert L. Causey Subject: Natural Kinds In a message posted 7/15, John McCarthy says that philosophers have recently introduced the concept of natural kind, and he suggests how this concept may be useful in AI. I think this deserves serious comment, both historical and substantive. The following is lengthy, but it may illustrate some general characteristics about the relationships between philosophy and AI. HISTORY In their messages, Ken Laws and others are correct -- the idea of natural kinds is not new. It is at least implicit in some Pre-Socratic Greek philosophy, and Aristotle extensively developed the idea and applied it in both philosophy and biology. Aristotle's conception is too "essentialist" to fit what McCarthy refers to. In the late 1600's John Locke developed an impressive empiricist analysis of natural kinds. Further developments were contributed in the 1800's in J. S. Mill's, _A_System_Of_Logic_. Mill also made important contributions to our understanding of inductive reasoning and scientific explanation; these are related to natural kinds. In our century a number of concepts of natural kinds have been proposed, ranging from strongly empiricist "cluster" approaches (which need NOT preclude expanding the cluster of attributes through the discovery of new knowledge, cf. McCarthy 7/17), to various modal analyses, to some intermediate approaches. Any of these analyses may have some value depending on the intended application, but the traditional notion of natural kinds has almost always been connected somehow with the idea of natural laws. SUBSTANTIVE ISSUES 1. Whatever one's favorite analysis might be, it is important to distinguish between a NATURAL kind (e.g., the compound silicon dioxide, with nomologically determined physical and chemical attributes), and a functional concept like "chair". There is generally not a simple one-to-one correspondence between our functional classifications of objects and the classification systems that are developed in the natural sciences. This is true in spite of the fact that we can learn to recognize sand, penguins, and chairs. But things are not always so simple - Suppose that Rip van Winkle learns in 1940 to recognize at sight a 1940-style adding machine; he then sleeps for 47 years. Upon waking in 1987 he probably would not recognize at sight what a thin, wallet calculator is. Functional classifications are useful, but we should not assume that they are generated and processed in the same ways as natural classifications. In particular, since functional classifications often involve an abstract understanding of complex behavioral dispositions, they are particularly hard to learn once one gets beyond simple things like chairs and tables. 2. Even discovering the classic examples of NATURAL kinds (like the classification of the chemical elements) can be a long and difficult process. It requires numerous inductive generalizations to confirm that the attributes in a certain Set of attributes each apply to gold, and that the attributes in some other Set of attributes apply to iodine, etc. We further recognize that our KNOWLEDGE of what are the elements of these Sets of attributes grows with the general growth of our scientific knowledge. Also, we need not always use the same set of attributes for IDENTIFICATION of instances of a natural kind. Most of this goes back to Locke, and philosophers have long recognized the connection between induction and classification; Carnap, Hempel, Goodman, and others, have sharpened some of the issues during the last 50 years. 3. Now, getting back to McCarthy's suggestion -- in his second message (7/17) he writes: "...for a child to presume a natural kind on hearing a word or seeing an object is advantageous, and it will also be advantageous to built (sic) AI systems with this presumption." His 7/15 message says, "When an object is named, the system should generate a gensym, e.g., GOO137. To this symbol should be attached the name and what the system is to remember about the instance." This is an interesting suggestion, but it prompts some comments and questions: i) Assuming that children do begin to presume natural kinds at some stage of development, what inductive processes are they using, what biologically determined constraints are affecting these processes, and what prior acquired knowledge is directing their inductions. These are interesting psychological questions. But, depending on our applications, we may not even want to build robots that emulate young children. We can attach a name to a gensym, but it is not at all easy to decide "...what the system is to remember about the instance," or to specify how it is to process all of the stuff it generates in this manner. ii) Children receive much corrective feedback from other people; how much feedback will we be willing or able to give to the "maturing" robots? Will the more mature robots help train the naive ones? iii) Given that classification does involve complex inductive reasoning, we need to learn a lot more about how to implement effective inductive procedures, where "induction" is understood very broadly. iv) If the AI systems (robots, etc.) are to learn, and reason with, functional concepts, then things get even more complex. Ability to make abstractions and perform complex analogical reasoning will be required. In my judgment, we (humans) still have a lot to learn just about the representation of functional knowledge. If my Rip van Winkle story seems farfetched, here is a true story. I know a person who is familiar with the appearance and use of 5 1/4 inch floppy diskettes. Upon first seeing a 3.5 inch mini-diskette, she had no idea what it was until its function was described. Knowledge of diskettes can extend to tracks, sectors, etc. The concept of natural kinds is relatively simple (though often difficult to apply); functional concepts and their relations with physical structures are harder subjects. ------------------------------ Date: 18 Jul 87 2315 PDT From: John McCarthy Subject: re: [Robert L. Causey : Natural Kinds] [In reply to message from AI.CAUSEY@R20.UTEXAS.EDU sent Sat 18 Jul 87.] I agree with Bob Causey's comments and agree that the open questions he lists are unsolved and important. I have one caveat. The distinction between nomological and functional kinds exists in sufficiently elaborate mental structures, but I don't think that under 2 year olds make the distinction, i.e. have different mechanisms for learning them. For this reason, it is an open question whether it should be a primary distinction for robots. In a small child's world, chairs are distinguished from other objects by appearance, not by function. Evidence: a child doesn't refer to different appearing objects on which he can also sit as chairs. Concession: there may be such a category "sittable" in "mentalese", and languages with such categories might be as easily learnable as English. What saves the child from having to make the distinction between kinds of kinds at an early age is that so many of the kinds in his life are distinguishable from each other in many ways. The child might indeed be fooled by the different generations of calculator, but usually he's lucky. I hope to comment later on how robots should be programmed to identify and use kinds. ------------------------------ End of AIList Digest ******************** 19-Jul-87 22:04:08-PDT,19848;000000000000 Mail-From: LAWS created at 19-Jul-87 21:59:00 Date: Sun 19 Jul 1987 21:52-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #185 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 20 Jul 1987 Volume 5 : Issue 185 Today's Topics: Perception - Seeing-Eye Robots, Philosophy - Searchability in Humans vs. Machines ---------------------------------------------------------------------- Date: 16 Jul 87 17:54:49 GMT From: ihnp4!homxb!houdi!marty1@ucbvax.Berkeley.EDU (M.BRILLIANT) Subject: Seeing-Eye robots Suppose one wanted to build a robot that does what a Seeing-Eye dog does (that is, helping a blind person to get around), but communicates in the blind person's own language instead of by pushing and pulling. Clearly this robot does not have to imitate a human being. But it does have to recognize objects and associate them with the names that humans use for them. It also has to interpret certain situations in its owner's terms: for instance, walking in one direction leads to danger, and walking in another direction leads to the goal. What problems will have to be solved to build such a robot? Will its hypothetical designers have to deal with the problem of mere recognition, or the deeper problem of grounding symbols in meaning? Could it be built by hardwiring sensors to a top-down symbolic processor, or would it require a hybrid processor? M. B. Brilliant Marty AT&T-BL HO 3D-520 (201)-949-1858 Holmdel, NJ 07733 ihnp4!houdi!marty1 ------------------------------ Date: 17 Jul 87 19:31:00 GMT From: ihnp4!inuxc!iuvax!merrill@ucbvax.Berkeley.EDU Subject: Re: Seeing-Eye robots In comp.ai, marty1@houdi (M.B. Brilliant) writes: > Suppose one wanted to build a robot that does what a Seeing-Eye dog > does (that is, helping a blind person to get around), but communicates > in the blind person's own language instead of by pushing and pulling. > [Commentary on some of the essential properties of the robot.] > What problems will have to be solved to build such a robot? Will its > hypothetical designers have to deal with the problem of mere > recognition, or the deeper problem of grounding symbols in meaning? > Could it be built by hardwiring sensors to a top-down symbolic > processor, or would it require a hybrid processor? I seriously doubt that recognition itself would be adequate. As Brilliant observes, one of the functions that the robot must perform is the detection of "danger to its master." Consider the problem of crossing a street. Is it enough to recognize cars (and trucks, and motorcycles, and other already--known objects?) No. The robodog has to generalize beyond simply cars and trucks and busses, since their shapes change, to "things travelling along this stretch of road {and what's a stretch of road?} which are a) moving {and what does it mean to move?} b) fast {and what is fast? Why, fast enough to be dangerous...which begs the question} c) in this direction." At this point, I think that we have exceeded the bounds of recognition and entered a realm where "judgement" is required, but, if not, I imagine that I can probably extend this situation to meet most specific objections. (I assume that the blind woman needs to cross roads without undue delay. Traffic lights don't eliminate these problems, since the robodog must "recognize" drivers who are turning, some of whow would be safe, since they're either stopped or slow--moving, but some of whom (at least, here in Bloomington) would run *any* pedestrian down. !-)) BTW: I like this example very much. It raises quite nicely the underlying issue in the symbol grounding problem discussion without using the terminology that many of the readers of comp.ai seem to have objected to. Congratulations, Mr. Brilliant! John Merrill merrill@iuvax.cs.indiana.edu UUCP:seismo!iuvax!merrill Dept. of Comp. Sci. Lindley Hall 101 Indiana University Bloomington, Ind. 47405 ------------------------------ Date: 14 Jul 87 21:21:33 GMT From: berke@locus.ucla.edu Subject: An Unsearchable Problem of Elementary Human Behavior An Unsearchable Problem of Elementary Human Behavior Peter Berke UCLA Computer Science Department The Artificial Intelligence assumption that all human behavior can eventually be mimicked by computer behavior has been stated in various ways. Since Newell stated his Problem Space Hypothesis in 1980, it has taken on a clearer, and thus, more refutable form. Newell stated his hypothesis thus: "The fundamental organizational unit of all human goal-oriented symbolic activity is the problem space." - Newell, 1980. In the 1980 work, Newell says his claim "hedges on whether all cognitive activity is symbolic." Laird, Rosenbloom, and Newell (1985) ignore this hedge and the qualification "goal- oriented symbolic" when they propose: "Our approach to developing a general learning mechanism is based on the hypothesis that all complex behavior - which includes behavior concerned with learning - occurs as search in problem spaces." They reference Newell (1980), but their claim is larger than Newell's original claim. The purpose of this note is to show that, to be true, Newell's hypothesis must be taken to mean just that goal-search in a state-space is a formalism that is equivalent to computing. Then Newell's Problem Space Hypothesis is simply a true theorem. The reader is invited to sketch a proof of the mutual simulatability of Turing computation and a process of goal-search in a state space. Such a proof has been constructed for every other prospective universal formalism, e.g., lambda calculus, recursive function theory, and Post tag systems. That such universal formalisms are equivalent in this sense led Church (1936, footnote 3) to speculate that human calculating activity can be given no more general a characterization. But human behavior is not restricted to calculating activity (though it seems that at least some human behavior is calculating). If the Problem Space Hypothesis is taken to be a stronger statement, that is, as a statement about human behavior rather than about the formalism of goal-search in a state-space, then I claim that the following counter-example shows it to be false. Understanding a name is an inherently unsearchable problem; It cannot be represented as search in a state or problem space. Well, it can be so represented, but then it is not the same problem. In searching our states for our goal we are solving a different problem than the original one. To understand that understanding is (or how it can be) inherently unsearchable, it is necessary to distinguish between ambiguity and equivocacy. At first the distinction seems contrived, but it is required by the assumption that there are discrete objects called 'names' that have discrete meaning (some other associated object or objects, see Church 1986, Berke 1987). An equivocal word/image has more than one clear meaning, an ambiguous word/image has none. What is usually meant by the phrase "lexical ambiguity" is semantic equivocacy. Equivocacy occurs even in formal languages and systems, though in setting up a formal system one aims to avoid equivocacy. For example, an expression in a computer language may be equivocal ("of equal voices"), such as: 'IF a THEN IF b THEN c ELSE d'. The whole expression is equivocal depending on which 'IF' the 'ELSE' is paired with. In this case there are two clear meanings, one for each choice of 'IF'. On the other hand, 'ELSE' taken in isolation, is ambiguous ("like both"), it's meaning is not one or many alternatives, but it is like all of them. [The reader, especially one who may claim that 'ELSE' has no meaning in isolation, may find it valuable to pause at this point to write down what 'ELSE' means. Several good attempts can be generated in very little time, especially with the aid of a dictionary.] Resolving equivocacy can be represented as search in a state space; it may very well BE search in a state space. Resolving ambiguity cannot be represented as search in a state space. Resolving environmental ambiguity is the problem-formulation stage of decision making; resolving objective ambiguity is the object-recognition phase of perception. The difference between ambiguity and equivocacy is a reason why object-recognition and problem-formulation are difficult programming and management problems, only iteratively approximable by computation or rational thought. A state space is, by definition, equivocal rather than ambiguous. If we confuse ambiguity with equivocacy, ambiguity resolution may seem like search in goal space, but this ignores the process of reducing an ambiguous situation to an equivocal one much the way Turing (1936) consciously ignores the transition of a light switch from OFF to ON. A digital process can approximate an analog process yet we distinguish the digital process from the analog one. Similarly, an equivocal problem can approximate an ambiguous problem, but the approximating problem differs from the approximated one. Even if a bank of mini-switches can simulate a larger light switch moving from OFF to ON, we don't evade the problem of switch transition, we push it "down" a level, and then ignore it. Even if we can simulate an ambiguity by a host of equivocacies, we don't thereby remove ambiguity, we push it "down" a level, and then ignore it. Ambiguity resolution cannot be accomplished by goal-search in a state space. At best it can be pushed down some levels. Ambiguity must still be resolved at the lower levels. It doesn't just go away; ambiguity resolution is the process of it going away. Representation may require ambiguity resolution, so the general problem of representing something (e.g., problem solving, understanding a name) as goal-search in a state space can not be represented as goal-search in a state space. This leads me to suspect what may be a stronger result: "Representing something" in a given formalism cannot be represented in that formalism. For example, "representing a thought in words," that is, expression, cannot be represented in words. "What it is to be a word" cannot be expressed in words. Thus there can be no definition of 'word' nor then of 'language'. Understanding a word, if it relies on some representation of "what it is to be a word" in words, cannot be represented in words. The meaning of a word is in this way precluded from being (or being adequately represented by) other words. This agrees with our daily observations that "the meaning of a word," in a dictionary is incomplete. Not all words need be impossible to completely define, just some of them for this argument to hold. It also agrees with Church's 1950 arguments on the contradictions inherent in taking words to be the meaning of other words. If understanding cannot be represented in words, it can never be well-defined and cannot be programmed. In programming, we can and must ignore the low-level process of bit-recognition because it is, and must be, implemented in hardware. Similarly, hardware must process ambiguities into equivocacies for subsequent "logical" processing. We are thus precluded from saying how understanding works, but that does not preclude us from understanding. Understanding a word can be learned as demonstrated by humans daily. Thus learning is not exhausted by any (word-expressed) formalism. One example of a formalism that does not exhaust learning behavior is computation as defined (put into words) by Turing. Another is goal-search in a state-space as defined (put into words) by Newell. References: Berke, P., "Naming and Knowledge: Implications of Church's Arguments about Knowledge Representation," in revision for publication,1987. Church, A., An Unsolvable Problem of Elementary Number Theory (Presented to the American Mathematical Society, April 19, 1935), Journal of Symbolic Logic, 1936. Church, A., "On Carnap's Analysis of Statements of Assertion and Belief," Analysis, 10:5, pp. 97-99, April, 1950. Church, A., "Intensionality and the Paradox of the Name Relation," Journal of Symbolic Logic, 1986. Laird, J.E., P.S. Rosenbloom, and A. Newell, "Towards Chunking as a General Learning Mechanism," CMU-CS-85-110. Newell, A. "Reasoning, Problem Solving, and Decision Processes: The problem space as a Fundamental Category. Chapter 35 in R. Nickerson (Ed.), Attention and Performance VIII. Erlbaum, 1980. Turing, A.M., On Computable numbers, with an application to the Entscheidungsproblem. Proceedings of the London Mathematical Society 42-2 (1936-7), 230-265; Correction, ibid., 43 (1937), 544-546. ------------------------------ Date: 16 Jul 87 09:23:07 GMT From: mcvax!botter!roelw@seismo.css.gov (Roel Wieringa) Subject: Berke's Unsearchable Problem In article 512 of comp.ai Peter Berke says that 1. Newell's hypothesis that all human goal-oriented symbolic activity is searching through a problem-space must be taken to mean that human goal-oriented symbolic activity is equivalent to computing, i.e. that it equivalent (mutually simulatable) to a process executed by a Turing machine; 2. but human behavior is not restricted to computing, the process of understanding an ambiguous word (one having 0 meanings, as opposed to an equivocal word, which has more than 1 meanings) being a case in point. Resolving equivocality can be done by searching a problem space; ambiguity cannot be so resolved. If 1 is correct (which requires a proof, as Berke says), then if 2 is correct, we can conclude that not all human behavior is searching through a problem space; the further conclusion then follows that classical AI (using computers and algorithms to reach its goal) cannot reach the goal of implementing human behavior as search through a state space. There are two problems I have with this argument. First, barring a quibble about the choice of the terms "ambiguity" and "equivocality", it seems to me that ambiguity as defined by Berke is really meaninglessness. I assume he does not mean that part of the surplus capacity of humans over machines is that humans can resolve meaninglessness whereas machines cannot, so Berke has not said what he wants to say. Second, the argument applies to classical AI. If one wishes to show that "machines cannot do everything that humans can do," one should find an argument which applies to connection machines, Boltzmann machines, etc. as well. Supposing for the sake of the argument that it is important to show that there is an essential difference between man and machine, I offer the following as an argument which avoids these problems. 1. Let us call a machine any system which is described by a state evolution function (if it has a continuous state space) or a state transition function (discrete state space). 2. Let us call a description explicit if (a) it is communicable to an arbitrary group of people who know the language in which the description is stated, (b) it is context-independent, i.e. mentions all relevant aspects of the system and its environment to be able to apply it, (c) describes a repeatable process, i.e. whenever the same state occurs, then from that point on the same input sequence will lead to the same output sequence, where "same" is defined as "described by the explicit description as an instance of an input (output) sequence." Laws of nature which describe how a natural process evolves, computer programs, and radio wiring diagrams are explicit descriptions. Now, obviously a machine is an explicitly described system. The essential difference between man and machine I propose is that man possesses the ability to explicate whereas machines do not. The *ability* to explicate is defined as the ability to produce an explicit description of a range of situations which (i.e. the range is) not described explicitly. In principle, one can build a machine which produces explicit descriptions of, say, objects on a conveyor belt. But the set of kinds of objects on the belt would then have to be explicitly described in advance, or at least it would in principle be explicitly describable, even though the description would be large, or difficult to find. the reason for this is that a machine is an explicitly described system, so that, among others, the set of possible inputs is explicitly described. On the other hand, a human being in principle can produce reasonably explicit descriptions of a class of systems which has no sharp boundaries. I think it is this capability which Berke means when he says that human beings can disambiguate whereas algorithmic processes cannot. If the set of inputs to an explication process carried out by a human being is itself not explicitly describable, then humans have a capability which machines don't have. A weak point in this argument is that human beings usually have a hard time in producing totally explicit descriptions; this is why programming is so diffcult. Hence, the qualification "reasonably explicit" above. This does not invalidate the comparison with machines, for a machine built to produce reasonably explicit descriptions would still be an explicitly described system, so that the sets of inputs and outputs would be explicitly described (in particular, the reasonableness of the explicitness of its output would be explicitly described as well). A second argument deriving from the concepts of machine and explicitness focuses on the three components of the concept of explicitness. Suppose that an explication process executed by a human being were explicitly describable. 1. Then it must be communicable; in particular the initial state must be communicable; but this seems one of the most incommunicable mental states there is. 2. It must be context-independent; but especially the initial stage of an explication process seems to be the most context-sensitive process there is. 3. It must be repeatable; but put the same person in the same situation (assuming that we can obliterate the memory of the previous explication of that situation) or put identical twins in the same situation, and we are likely to get different explicit descriptions of that situation. Note that these arguments do not use the concept of ambiguity as defined by Berke and, if valid, apply to any machine, including connection machines. Note also that they are not *proofs*. If they were, they would be explicit descriptions of the relation between a number of propositions, and this would contradict the claim that the explication process has very vague beginnings. Roel Wieringa ------------------------------ End of AIList Digest ******************** 21-Jul-87 22:36:36-PDT,15166;000000000000 Mail-From: LAWS created at 21-Jul-87 22:28:10 Date: Tue 21 Jul 1987 22:25-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #186 To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 22 Jul 1987 Volume 5 : Issue 186 Today's Topics: Queries - AI Application for DBase of New Chemical Substances & Software Reuse, Science Fiction - Immortality via Computer, Techniques - Garbage Collection Suppression, Philosophy - Natural Kinds, Courses - Philosophy Courses on Artificial Intelligence & Logic and Computability, AI and Formal Learning Theory ---------------------------------------------------------------------- Date: 20 Jul 1987 1220-EDT From: Holger Sommer Subject: AI/Expert system application for DBase of New Chemical Substances I am involved in an EPA/NIOSH sponsered project with the title : "Engineering and Toxic Characterisation Studies and Development of Unit Operations Predictive Models for New Chemicals" This project intends to develop an intelligent database and predictive models to help EPA in the evaluation of premanufacture notices. The research is directed and conducted to establish data base model for use in predicting workplace releases and exposures resulting from manufacturing , processing , use or disposal of new chemical substances. The main activities of this project are : * Formulation of conceptual data base models for filtration and drying unit operation * Assessment and characterization of worker exposure in manufacturing plants and pilot plants * Incorporation of sampling data and other relevant information into the data base framework * Development and validation of computerized predictive models for assessment of workplace releases and exposures. What we try to accomplish in this project is to automate the evaluation process for premanufacture notices and provide a systematic data base to assist in this evaluation. My questions to the AI-list audience are : 1) Are there projects underway with similar content ( not particular related to chemicals but other domains ) ? 2) We need information about existing data base programs which interface with predictive models. We are looking for a flexible programming tool to accomplish the above described assignments. Thank you for any information I will receive through this network. Please send responses to : H.T. Sommer .... Sommer@c.cs.cmu.edu ------------------------------ Date: 18 Jul 87 15:17:05 GMT From: cbmvax!vu-vlsi!ge-mc3i!sterritt@RUTGERS.EDU (Chris Sterritt) Subject: Re: Software Reuse (short title) Hello, I've been following the discussion of this avidly, but am new to the programming languages (?) ML, SML, and LML. Could someone (ideally mail me directly so as not to clog the net!) send me information on these langauges, so that I might find out more? Along the ideas of the discussion, if I remember my Computability theory correctly -- doesn't it make some sense that to show an algorithm (either computable or to prove it) you need to give an almost algorithmic description, as in an inductive proof? So isn't this what Lisp is (I'm a lisp hacker at work). I'd think that Church's Lambda Calculus would shed some light on this discussion, as I believe that that was what he was trying to do with the calculus. Generally, I agree that to specify an algorithm IN ENOUGH DETAIL, you will probably wind up writing at least as much information down as the code itself. I think that 'Requirements' as we define them in 'Software Engineering' presume a *lot* of human intelligence. Any comments? Chris Sterritt ------------------------------ Date: 20 Jul 87 13:21:39 GMT From: sunybcs!rapaport@RUTGERS.EDU (William J. Rapaport) Reply-to: sunybcs!rapaport@RUTGERS.EDU (William J. Rapaport) Subject: Re: Immortality via Computer In article <8707200504.AA05729@ucbvax.Berkeley.EDU> MNORTON@rca.COM writes: > >Concerning the AP story on attaining immortality via computers, readers >of AIList intrested in thinking more about this may wish to read ... ... or Justin Leiber's _Beyond Rejection_. Leiber is a philosopher and also the son of SF writer Fritz Leiber. The novel is about a society in which brain tapes are made and installed in new bodies; the minds tend to reject the bodies. ------------------------------ Date: Mon, 20 Jul 87 21:43:00 EDT From: Chester@UDEL.EDU Subject: Re: Garbage Collection Suppression The direct way to avoid garbage collection in lisp is to define your own `cons' function that prefers to get cell pairs from an `available list', calling the regular `cons' only when the `available list' is empty. A `reclaim' function that puts cell pairs on the `available list' (using `rplacd') will be needed also. See any book on data structures. The technique can be used for cell pairs and gensym atoms, if needed, but in my experience, not with strings or numbers. String manipulations can usually be avoided, but a program that crunches a lot of numbers cannot avoid consuming memory and eventually triggering garbage collection (at least in VAX lisp). I wish there were some way for a user to reclaim numbers so that they could be reused as cell pairs can. If so, I could write all my lisp programs so that they don't need to garbage collect. It would also be nice to have a built-in `reclaim' function that would work in conjunction with the built-in `cons'; it would be dangerous for novices, but handy for the experienced. By the way, recursion in itself doesn't cause garbage collection; VAX lisp is smart enough to reclaim the memory used for the function stack automatically. Daniel Chester chester@dewey.udel.edu ------------------------------ Date: 21 Jul 87 17:05:53 GMT From: rlw@philabs.philips.com (Richard Wexelblat) Reply-to: rlw@philabs.philips.com (Richard Wexelblat) Subject: Re: Natural Kinds It is amusing and instructive to study and speculate on children's language and conceptualization. (Wow! That construct's almost Swiftean!) For those who would read further in this domain, I recommend: Brown, Roger A First Language -- The Early Stages Harvard Univ. Press, 1973 MacNamara, John Names for Things -- A Study of Human Learning MIT Press, 1984 ------------------------------ Date: 21 Jul 87 16:56:08 GMT From: rlw@philabs.philips.com (Richard Wexelblat) Reply-to: rlw@briar.philips.com (Richard Wexelblat) Subject: Re: Natural Kinds In article <8707161942.AA13065@nrl-css.ARPA> mclean@NRL-CSS.ARPA (John McLean) writes: >However, I think the issue being raised about recognizing penguins, >chairs, etc. goes back to Wittgenstein's _Philosophical_Investigations_: Actually, the particular section chosen is a bit too terse. Here is more context: Consider, for example the proceedings that we call `games.' I mean board- games, card-games, ball-games, Olympic games, and so on. What is common to them all?--Don't say: ``There must be something common, or they would not be called `games' ''--but look and see whether there is anything common to all. --For if you look at them you will not see something that is common to all, but similarities, relationships, and a whole series of them at that ... a complicated network of similarities overlapping and criss-crossing; sometimes overall similarities, sometimes similarities of detail. I can think of no better expression to characterize these similarities than ``family resemblances''; for the various resemblances between the members of a family: build, features, colour of eyes, gait, temperament, etc. etc. overlap and criss-cross in the same way.--And I shall say: `games' form a family. * * * This sort of argument came up in a project on conceptual design tools a few years ago in attempting to answer the question: ``What is a design and how do you know when you have one?'' We attempted to answer the question and got into the question of subjective classifications of architecture. What is a ``ranch'' or ``colonial'' house? If you can get a definition that will satisfy a homebuyer, you are in the wrong business. * * * Gratis, here are two amusing epigrams from W's Notebooks, 1914-1916: There can never be surprises in logic. ~~~~~ One of the most difficult of the philosopher's tasks is to find out where the shoe pinches. ------------------------------ Date: 17 Jul 1987 1505-EDT From: Clark Glymour Subject: Philosophy Courses on Artificial Intelligence SEMINAR IN LOGIC AND COMPUTABILITY: ARTIFICIAL INTELLIGENCE AND FORMAL LEARNING THEORY - Offered by: Department of Philosophy, Carnegie-Mellon University - Instructor: Kevin T. Kelly - Graduate Listing: 80-812 - Undergraduate Listing: 80-510 - Place: Baker Hall 131-A - Time: Wednesday, 1:30 to 4:30 (but probably not the full period). - Intended Audience: Graduate students and sophisticated undergraduates interested in inductive methods, the philosophy of science, mathematical logic, statistics, computer science, artificial intelligence, and cognitive science. - Prerequisites: A good working knowledge of mathematical logic and computation theory. - Course Focus: Convergent realism is the philosophical thesis that the point of inquiry is to converge (in some sense) to the truth (or to something like it). Formal learning theory is a growing body of precise results concerning the possible circumstances under which this ideal is attainable. The basic idea was developed by Hilary Putnam in the early 1960's, and was extended to questions in theoretical linguistics by E. Mark Gold. The main text of the seminar will be Osherson and Weinstein's recent book Systems that Learn. But we will also examine more recent efforts by Osherson, Weinstein, Glymour and Kelly to apply the theory to the inductive inference of theories expressed in logical languages. From this general standpoint, we will move to more detailed projects such as the recent results of Valiant, Pitt, and Kearns on polynomial learnabilitly. Finally, we will examine the extent to which formal learning theory can assist in the demonstrable improvement of learning systems published in the A.I. machine learning literature. There is ample opportunity to break new ground here. Thesis topics abound. - Course Format: Several introductory lectures, Seminar reports, and a novel research project. PROBABILITY AND ARTIFICIAL INTELLIGENCE - Offered by: Department of Philosophy, Carnegie-Mellon University - Instructor: Kevin T. Kelly - Graduate Course Number: 80-312 - Undergraduate Course Number: 80-811 - Place: Porter Hall, 126-B - Time: Tuesday, Thursday, 3:00-4:20 - Intended Audience: Graduate students and sophisticated undergraduates interested in inductive methods, the philosophy of science, mathematical logic, statistics, computer science, artificial intelligence, and cognitive science. - Prerequisites: Familiarity with mathematical logic, computation, and probability theory - Course Focus: There are several ways in which the combined system of a rational agent and its environment can be stochastic. The agent's hypotheses may make claims about probabilities, the agent's environment may be stochastic, and the agent itself may be stochastic, in any combination. In this course, we will examine efforts to study computational agents in each of these situations. The aim will be to assess particular computational proposals from the point of view of logic and probability theory. Example topics are Bayesian systems, Dempster-Shafer theory, medical expert systems, computationally tractable learnability, automated linear causal modelling, and Osherson and Weinstein's results concerning limitations on effective Bayesians. - Course Format: The grade will be based on frequent exercises and possibly a final project. There will be no examinations if the class keeps up with the material. ------------------------------ Date: 17 Jul 87 16:54:45 EDT From: Terina.Jett@b.gp.cs.cmu.edu Subject: Seminar - Logic and Computability, AI and Formal Learning Theory SEMINAR IN LOGIC AND COMPUTABILITY ARTIFICIAL INTELLIGENCE AND FORMAL LEARNING THEORY Offered by: Department of Philosophy Instructor: Kevin T. Kelly Grad Listing: 80-510 Undergrad Listing: 80-510 Place: Baker Hall 131-A Time: Wed, 1:30 - 4:30 Intended Audience: Graduate students and sophisticated undergraduates interested in inductive methods, the philosophy of science, mathematical logic, statistics, computer science, artificial intelligence, and cogni- tive science. Prerequisites: A good working knowledge of mathematical logic and comp- utation theory. Course Focus: Convergent realism is the philosophickal thesis that the point of inquiry is to converge (in some sense) to the truth (or to something like it). Formal learning theory is a growing body of precise results concerning the possible circumstances under which this ideal is attainable. The basic idea was developed by Hilary Putnam in the early 1960's, and was extended to questions in theoretical linguistics by E. Mark Gold. The main text fo the seminar will be Osherson and Weinstein's recent book Systems That Learn. But we will also examine more recent efforts by Osherson, Weinstein, Glymour and Kelly to apply the theory to the inductive inference of theories expressed in logical languages. From this general standpoint, we will move to more detailed projects such as the recent results of Valiant, Pitt, and Kearns on polynomials learn- abilitly. Finally, we will examine the extent to which formal learning theory can assist in the demonstrable improvement of learning systems published in the A.I. machine learning literature. There is ample opportunity to break new ground here. Thesis topics abound. Course Format: Serveral introductory lectures, Seminar reports, and a novel research project. ------------------------------ End of AIList Digest ******************** 26-Jul-87 23:42:07-PDT,18922;000000000000 Mail-From: LAWS created at 26-Jul-87 23:32:49 Date: Sun 26 Jul 1987 23:31-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #187 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 27 Jul 1987 Volume 5 : Issue 187 Today's Topics: Journal Issue - Planning (Int. J. for AI in Engineering), Seminar - Abstraction in Knowledge-Based Systems (MCC), Course - Probability and AI (CMU), Conferences - CD-ROM & 7th Distributed Computing Systems & R&D in Information Retrieval & International Neural Network Society ---------------------------------------------------------------------- Date: Fri, 24 Jul 87 09:29:43 EDT From: sriram@ATHENA.MIT.EDU Subject: Journal Issue - Planning (Int. J. for AI in Engineering) INTERNATIONAL JOURNAL FOR AI IN ENGINEERING SPECIAL ISSUE ON PLANNING APRIL 1988 The April 1988 issue of the International Journal for AI in Engineering will be dedicated to Planning. The guest editors for this issue are: Prof. Chris Hendrickson, Dept. of Civil Engineering, C-MU, Pittsburgh, PA 15213 (hendrickson@cive.ri.cmu.edu) and Mrs Julie Gadsden, Admiralty Research Establishment, Procurement Executive, XCC5.2, Portsdown, Portsmouth, Hants PO6 4AA, UK. Papers in all areas of engineering, as related to planning, are solicited. Each paper should not exceed 10,000 words (roughly 30 doubly spaced pages), including figures. The deadline for submission is September 1, 1987. Please send the papers to either of the guest editors. Sriram & McCallum (Editors) ------------------------------ Date: Fri 24 Jul 87 11:59:41-CDT From: Betti Bunce Subject: Seminar - Abstraction in Knowledge-Based Systems (MCC) All interested parties are invited to attend the following: TALK BY: B. Chandrasekaran Laboratory for AI Research Department of Computer and Information Science The Ohio State University Columbus, OH 43210 DATE: August 5, 1987 TIME: 10:00 a.m. WHERE: MCC Auditorium 3500 West Balcones Center Drive CONTACTS: Charles Petrie - MCC Ben Kuipers - UT TITLE: THE GENERIC TASK TOOLKIT FOR KNOWLEDGE-BASED SYSTEMS: BUILDING BLOCKS AT THE ``RIGHT'' LEVEL OF ABSTRACTION ABSTRACT: The first part to the talk is a critique of the level of abstraction of much of the current discussion on knowledge-based systems. It will be argued that the discussion at the level of rules-logic-frames-networks is the ``civil engineering'' level, and there is a need for a level of abstraction that corresponds to what the discipline of architecture does for construction of buildings. The constructs in architecture, viewed as a language of habitable spaces, can be implemented using the constructs of civil engineering, but are not reducible to them. Similarly, level of abstraction that we advocate is the language of generic tasks, types of knowledge and control regimes. In the second part of the talk, I will outline the elements of a framework at this level of abstraction for expert system design that we have been developing in our research group over the last several years. Complex knowledge-based reasoning tasks can often be decomposed into a number of generic tasks each with associated types of knowledge and family of control regimes. At different stages in reasoning, the system will typically engage in one of the tasks, depending upon the knowledge available and the state of problem solving. The advantages of this point of view are manifold: (i) Since typically the generic tasks are at a much higher level of abstraction than those associated with first generation expert system languages, knowledge can be represented directly at the level appropriate to the information processing task. (ii) Since each of the generic tasks has an appropriate control regime, problem solving behavior may be more perspicuously encoded. (iii) Because of a richer generic vocabulary in terms of which knowledge and control are represented, explanation of problem solving behavior is also more perspicuous. We briefly describe six generic tasks that we have found very useful in our work on knowledge-based reasoning: classification, state abstraaction, knowledge-directed retrieval, object synthesis by plan selection and refinement, hypothesis matching, and assembly of compound hypotheses for abduction. Finally, we will describe how the above approach leads naturally to a new technology: a toolbox which helps one to build expert systems by using higher level building blocks. We will review the toolbox, and outline what sorts of systems can be built using the toolbox, and what advantages accrue from this approach. ------------------------------ Date: 20 Jul 87 12:15:08 EDT From: Terina.Jett@b.gp.cs.cmu.edu Subject: Course - Probability and AI (CMU) PROBABILITY AND ARTIFICIAL INTELLIGENCE Offered by: Department of Philosophy, CMU Instructor: Kevin T. Kelly Grad Course No: 80-312 Undergrad Course No: 80-811 Place: Porter Hall, 126-B Time: Tuesday, Thursday, 3:00-4:00 Intended Audience: Graduate students ans sophisticated undergraduates interested in inductive methods, the philosophy of science, mathematical logic, statistics, computer science, artificial intelligence, and cognitive science. Prerequisites: Familiarity with mathematical logic, computation, and probability theory. Course Focus: There are several ways in which the combined system of a rational agent and its environment can be stochastic. The agent's hypotheses may make claims about probabilities, the agent's environment may be stochastic, and the agent itself may be stochastic, in any com- bination. In this course, we will examine efforts to study computational proposals from the point of view of logic and probability theory. Example topics are Bayesian systems, Dempster/Shafer theory, medical expert systems, computationally tractable learnability, automated linear causal modelling, and Osherson and Weinstein's results concerning limitations on effective Bayesians. Course Format: The grade will be based on frequent exercises and possibly a final project. There will be no examinations if the class keeps up with the material. ------------------------------ Date: Fri, 17 Jul 1987 14:58 CST From: Leff (Southern Methodist University) Subject: Conferences - CD-ROM & 7th Distributed Computing Systems AI at Upcoming Conferences CD-ROM Expo, New York City September 21-23 T-8 Using CD-RoM in Expert Systems H-1 Helping the Non-Expert Use CD-ROM, Artificial Intellgience and Expert Systems Seventh International Conference on Distributed Computing Systems Berlin (West), 21-25th September 1987 Thursday, 24 Sep 1987, 11.00-12.30 On the Application of AI in Decentralized Cnotrol: An Illustration by Mutual Exclusion 1987 International Conferenceon Parallel Processing Tutorial 10:30AM Dr. Benjamin W. Wah, Computers for Artificial Intelligence Processing. A Parallel Model and Architecture and Architecturee for Production Systems by A. O. Oshisanwo and P. P. Dasiewicz Parallel Link Resolution of Connection Graph Refutation and its Implementation by R. Loganantharaj (Logan) Combinators as Control Mechanisms in Multiprocessing Systems by D. L. Knox and C. T. Wright An AND-OR Parallel Execution System for Logic Program Evaluation by N. S. Woo and R. Sharma PESA I - A Parallel Architecture for Production Systems by F. Schreiner and G. Zimmermann A New Parallel Graph Reduciton Model and its Machine Architecture by M. Amamiya Parallel Garbage Collection on a Virtual Memory System by S. G. Abraham and J. H. Patel A Knowledge-Based Parallelization Tool in a Programming Environment by T. Brandes and M. Sommer A Heuristic Algorithm for Conflict Resolution Problem in Multistage Interconnection Networks by J. S. Deogun and Z. Fang Exploiting Locality of Reference in MIMD Parallel Symbolic Computation by Y. Eisenstadter and G. Q. McGuire, Jr. Efficient Image Template Matching on Hypercube SIMD Arrays by V. K. P. Kumar and V. Krishnan Practical Algorithms for Image Component Labeling on SIMD Mesh Connected Computers by R. E. Cypher, J. L. C. Sanz and L. Snyder A Parralel O(logN) algorithm for Finding Connected Components in Planar Images by A. Agrawal, L. Nekludova and W. LIM Large Scale Unification Using a Mesh-Connected Array of Hardware Unifiers by Shih and K. B. Irani On Source to Source Transformation of Sequential Logic Programs to AND- parallelism by A. K. Bansal and L. S. Sterling An Overlapping Unification Algorithm and its Hardware Implementation by W. T. Chen and K. R. Hseih Pipelined Evaluation of Conjunctive PRoblems by S. C. Sheu Analysis and Design of Parallel Aglortihms and Implementations for Some Image Processing Operations by M. Yasrebi, J. C. Browne and D. P. Agrawal parallel Image Processing on enhanced Arrays by V. K. P. Kumar and D. Reisis Parallel Pattern Clustering on a Multiprocessor with Orthogally Shared Memory by K. Hwang and D. Kim A General Purpose VLSI Array for Efficient Signal and Image Processing by S. Sastry and V. K. P. Kumar Computing the Two-Dimensional Discrete Fourier Transforma on the ASPEn Paralle Computer Architecture by A. L. Gorin, A. Silberger and L. Auslander ------------------------------ Date: Mon, 20 Jul 87 13:40:35 CDT From: Don Subject: Conference - R&D in Information Retrieval I have just received a travel grant for twenty or so stipends covering airfare from the National Science Foundation so that U.S. residents can attend the ACM/SIGIR International Conference on Research and Development in Information Retrieval, to be held in Grenoble, France on June 13-15, 1988. The conference will include the topics of retrieval system modeling, artificial intelligence and information retrieval, evaluation techniques, hardware developments for retrieval systems, natural language processing, database management and information retrieval, user interfaces, and advanced applications. Anyone interested in receiving a travel stipend should contact me. The deadline for applying for a travel stipend is March 1, 1988. Submission of papers (four copies of either a full paper of not more than 20-25 pages, or an extended abstract of about ten pages) with a complete author identification and an abstract of about one hundred words must be submitted by January 15, 1988 to: Professor Gerard Salton Department of Computer Science 4130 Upson Hall Cornell University Ithaca, NY 14853-7501 USA Final copy is due May 16, 1988, with acceptance notification coming by March 21, 1988. Don Kraft kraft@lsu.edu ------------------------------ Date: Tue, 21 Jul 87 09:39 EDT From: MIKE%BUCASA.BITNET@wiscvm.wisc.edu Subject: Conference - International Neural Network Society INTERNATIONAL NEURAL NETWORK SOCIETY 1988 ANNUAL MEETING September 6--10, 1988 Boston, Massachusetts The International Neural Network Society (INNS) is an association of scientists, engineers, students, and others seeking to learn about and advance our understanding of the modelling of behavioral and brain processes, and the application of neural modelling concepts to technological problems. The INNS invites all those interested in the exciting and rapidly expanding field of neural networks to attend its 1988 Annual Meeting. The planned conference program includes plenary lectures, symposia on selected topics, contributed oral and poster presentations, tutorials, commercial and publishing exhibits, a placement service for employers and educational institutions, government agency presentations, and social events. Individuals from fields as diverse as engineering, psychology, neuroscience, computer science, mathematics, and physics are now engaged in neural network research. This diversity is reflected in both the 1988 INNS Annual Meeting Advisory Committee and in the Editorial Board of the INNS journal, Neural Networks. In order to enhance the effectiveness of these multidisciplinary ventures and to inform a wide audience, organization of the INNS Annual Meeting will be carried out with the active participation of several professional societies. Meeting Advisory Committee includes: Demetri Psaltis---Meeting Chairman Larry Jackel---Program Chairman Gail Carpenter---Organizing Chairman Shun-ichi Amari James Anderson Maureen Caudill Walter Freeman Kunihiko Fukushima Lee Giles Stephen Grossberg Robert Hecht-Nielsen Teuvo Kohonen Christoph von der Malsburg Carver Mead Edward Posner David Rumelhart Terrence Sejnowski George Sperling Harold Szu Bernard Widrow CALL FOR ABSTRACTS: The INNS announces an open call for abstracts to be considered for oral or poster presentation at its 1988 Annual Meeting. Meeting topics include: --Vision and image processing --Speech and language understanding --Sensory-motor control and robotics --Pattern recognition --Associative learning --Self-organization --Cognitive information processing --Local circuit neurobiology --Analysis of network dynamics --Combinatorial optimization --Electronic and optical implementations --Neurocomputers --Applications Abstracts must be typed on the INNS abstract form in camera-ready format. An abstract form and instructions may be obtained by returning the enclosed request form to: Neural Networks, AT&T Bell Labs, Room 4G-323, Holmdel, NJ 07733 USA. In order to be considered for presentation at the INNS 1988 Annual Meeting, an abstract must be POSTMARKED NO LATER THAN March 31, 1988. Acceptance notifications will be mailed by June 30, 1988. An individual may make at most one oral presentation during the contributed paper sessions. Abstracts accepted for presentation at the Meeting will be published as a supplement to the INNS journal, Neural Networks. Published abstracts will be available to participants at the conference. ***** ABSTRACT DEADLINE: MARCH 31, 1988 ***** CONFERENCE SITE: The 1988 Annual Meeting of the International Neural Network Society will be held at the Park Plaza Hotel in downtown Boston. A block of rooms has been reserved for the INNS at the rate of $91 per night plus tax (single or double). Reservations may be made by contacting the hotel directly. Be sure to give the reference "Neural Networks". A one-night deposit will be requested. HOTEL RESERVATIONS: Boston Park Plaza Hotel "Neural Networks" 1 Park Plaza at Arlington Street Boston, MA 02117 USA (800) 225-2008 (continental U.S.) (800) 462-2022 (Massachusetts only) Telex 940107 INTERNATIONAL RESERVATIONS: Steigenberger, Utell International KLM Golden Tulip, British Airways REF: "Neural Networks" Please note that other nearby hotel accomodations are typically more expensive and may also sell out quickly. CONFERENCE REGISTRATION: To register for the 1988 INNS Annual Meeting, return the enclosed conference registration form, with registration fee; or contact: UNIGLOBE---Neural Networks 1988, 40 Washington Street, Wellesley Hills, MA 02181 USA, (800) 521-5144 or (617) 235-7500. The great interest and attention now being devoted to the field of neural networks promises to generate a large number of meeting participants. Conference room size and hotel accomodations are limited. Therefore early registration is strongly advised. For information about INNS membership, which includes a subscription to the INNS journal, Neural Networks, write: Dr. Harold Szu---INNS, NRL Code 5756, Washington, DC 20375-5000 USA, (202) 767-1493. ADVANCE REGISTRATION FEE SCHEDULE INNS Member Non-member Until March 31, 1988 $125 $170* Until July 31, 1988 $175 $220* Full-time student $50 $85* * Includes the option of electing one-year INNS membership and subscription to the INNS journal, Neural Networks, free of charge. The conference registration fee schedule has been set to cover abstract handling costs, the book of abstracts, a buffet dinner reception, coffee breaks, informational mailings, and administrative expenses. Anticipated financial support by government and corporate sponsors will cover additional basic meeting costs. Tutorials and other special programs will require payment of additional fees. STUDENTS AND VOLUNTEERS: Students are particularly welcome to join the INNS and to participate fully in its Annual Meeting. Reduced registration and membership rates are available for full-time students. In addition, financial support is anticipated for students and meeting volunteers. To apply, please enclose with the conference registration application a letter of request and a brief description of interests. -----ABSTRACT REQUEST FORM----- INTERNATIONAL NEURAL NETWORK SOCIETY 1988 ANNUAL MEETING September 6--10, 1988 Boston, Massachusetts Please send an abstract form and instructions to: Name: Address: Telephone(s): All abstracts must be submitted camera-ready, typed on the INNS abstract form and postmarked NO LATER THAN March 31, 1988. MAIL TO: Neural Networks AT&T Bell Labs Room 4G-323 Holmdel, NJ 07733 USA -----REQUEST FOR INFORMATION----- INTERNATIONAL NEURAL NETWORK SOCIETY 1988 ANNUAL MEETING September 6--10, 1988 Boston, Massachusetts Please send information on the following topics to: Name: Address: Telephone(s): ( ) Placement/Interview service ( ) Employer ( ) Educational institution ( ) Candidate ( ) Hotel accomodations ( ) Travel and discounted fares Discounts of up to 60% off coach fare can be obtained on conference travel booked through UNIGLOBE: (800) 521-5144 or (617) 235-7500. ( ) Volunteer and student programs ( ) Proposals for symposia and special programs ( ) Exhibits ( ) Commercial vendor ( ) Publisher ( ) Government agency ( ) Tutorials ( ) Press credentials ( ) INNS membership MAIL TO: Center for Adaptive Systems---INNS Boston University 111 Cummington Street, Room 244 Boston, Massachusetts 02215 USA ELECTRONIC MAIL TO: mike@bucasa.bu.edu ------------------------------ End of AIList Digest ******************** 26-Jul-87 23:45:47-PDT,16081;000000000000 Mail-From: LAWS created at 26-Jul-87 23:39:38 Date: Sun 26 Jul 1987 23:37-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #188 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 27 Jul 1987 Volume 5 : Issue 188 Today's Topics: Queries - Graphics/AI Bibliography & Blackboard Architectures in Prolog & VPExpert Parameters & Knowledge Representation in Sanskrit, Techniques - Garbage Collection Suppression, Philosophy - Natural Kinds & AI, Science, and Pseudo-Science ---------------------------------------------------------------------- Date: 22-JUL-1987 15:13:29 From: THOWARD%graphics.computer-science.manchester.ac.uk@Cs.Ucl.AC.UK Subject: Graphics-AI bibliography I am currently investigating what work has been done on connecting/integrating AI methods and computer graphics. I would be very grateful if anyone can send me any references, or bibliographies (or comments!) etc in this area. If there's enough interest, I will summarise responses. Thanks... ______________________________________________________________________________ - Toby Howard - Computer Graphics Unit, Department of Computer Science Manchester University, England, M13 9PL. Phone: 061 273 7121 x5429/5406 Janet: thoward@uk.ac.man.cs.cgu ARPA: thoward%cgu.cs.man.ac.uk@cs.ucl.ac.uk ------------------------------ Date: Wed 22 Jul 87 11:52:28-CDT From: OLIVIER J. WINGHART Subject: Blackboard architectures in Prolog I am looking for natural ways of implementing a blackboard architecture in Prolog. Has anyone already thought about this, and are there any papers that I could look at ? I would appreciate any pointer. Olivier cs.winghart@utexas.edu ------------------------------ Date: Fri, 24 Jul 87 18:00:05 EDT From: Brady@UDEL.EDU Subject: VPExpert Parameters The VPExpert manual says that data can be passed to a batch file, and that this is the only way to directly pass parameters to an external program. But when I try to do this, the system tells me the syntax of my call is wrong. I am sure my error is not in the call to the batch file itself, since I am able to call and execute a batch file that does not require parameters. Anyone out there using this shell who has figured out how to pass parameters to a batch file, please send me mail. I will post answers back to the net. Thank you. ///////// joe brady ------------------------------ Date: Thu, 23 Jul 87 10:54:47 PDT From: bwidlans%zodiac@ads.arpa (Bob Widlansky) Subject: Knowledge Representation in Sanskrit Recently, I read a short intriguing article in AI magazine about the First International Conference on Knowledge Representation and Inference in Sanskrit (held in Bangalore, India between December 20-22, 1986). Does anyone know where I can get a copy of the proceedings? If you do, please contact me at bwidlans@ads.ARPA Thank you, Bob Widlansky ------------------------------ Date: 22 Jul 87 14:28:51 GMT From: "J. A. \"Biep\" Durieux" Reply-to: "J. A. \"Biep\" Durieux" Subject: Re: Garbage Collection Suppression In article <8707202143.aa23792@Dewey.UDEL.EDU> Chester@UDEL.EDU writes: >The direct way to avoid garbage collection in lisp is to define your own `cons' >function that prefers to get cell pairs from an `available list' (...). Also handy in many cases (small functions like append, alist-functions, subst) is icons: (defun icons (a d cell) (cond ((and (eq (car cell) a) (eq (cdr cell) d)) cell) (t (cons a d)))) In this way whenever it turns out the new cells weren't really needed, the old ones are used again (as in (append x nil)). Be aware, however, that your copy-function may not work any more if it's defined as (subst nil nil x)! -- Biep. (biep@cs.vu.nl via mcvax) Never confound beauty with truth! ------------------------------ Date: Wed, 22 Jul 1987 10:43 EDT From: MINSKY%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: Natural Kinds (Re: AIList Digest V5 #186) About natural kinds. In "The Society of Mind", pp123-129, I propose a way to deal with Wittgenstein's problem of defining terms like "game"- or "chair". The basic idea was to probe further into what Wittgenstein was trying to do when he talked about "family resemblances" and tried to describe a game in terms of properties, the way one might treat members of a human family: build, features, colour of eyes, gait, temperament, etc. In my view, Wittgenstein missed the point because he focussed on "structure" only. What we have to do is also take into account the "function", "goal", or "intended use" of the definition. My trick is to catch the idea between two descriptions, structural and functional. Consider a chair, for example. STRUCTURE: A chair usually has a seat, back, and legs - but any of them can be changed in so many ways that it is hard to make a definition to catch them all. FUNCTION: A chair is intended to be used to keep one's bottom about 14 inches off the floor, to support one's back comfortably, and to provide space to bend the knees. If you understand BOTH of these, then you can make sense of that list of structural features - seat, back, and legs - and engage your other worldly knowledge to decide when a given object might serve well as a chair. This also helps us understand how to deal with "toy chair" and such matters. Is a toy chair a chair? The answer depends on what you want to use it for. It is a chair, for example, for a suitable toy person, or for reminding people of "real" chairs, or etc. In other words, we should not worship Wittgenstein's final defeat, in which he speaks about vague resemblances - and, in effect, gives up hope of dealing with such subjects logically. I suspect he simply wasn't ready to deal with intentions - because nothing comparable to Newell and Simon's GPS theory of goals, or McCarthy's meta-predicate (Want P) was yet available. I would appreciate comments, because I think this may be an important theory, and no one seems to have noticed it. I just noticed, myself, that I didn't mention Wittgenstein himself (on page 130) when discussiong the definition of "game". Apologies to his ghost. ------------------------------ Date: Wed, 22 Jul 87 12:40:58 EDT From: mckee%corwin.ccs.northeastern.edu@RELAY.CS.NET Subject: AI, science, and pseudo-science In AIlist Digest v5 #171, July 6, 1987, Don Norman wrote: > [Here's why] many of us otherwise friendly folks in the sciences that > neighbor AI [are] frustrated with AI's casual attitude toward theory: > AI is not a science and its practitioners are woefuly untutored in > scientific method." [ 15 lines deleted ] > AI worries a lot about methods and techniques, with many books and > articles devoted to these issues. But by methods and techniques I > mean such topics as the representation of knowledge, logic, > programming, control structures, etc. None of this method includes > anything about content. And there is the flaw: nobody in the field of > Artificial Intelligence speaks of what it means to study intelligence, > of what scientific methods are appropriate, what emprical methods are > relevant, what theories mean, and how they are to be tested. All the > other sciences worry a lot about these issues, about methodology, > about the meaning of theory and what the appropriate data collection > methods might be. AI is not a science in this sense of the word. [ 22 more lines deleted ] I think he's found an issue of critical importance here, so I'm going to pull it out of context even further and repeat it again: "nobody in the field of Artificial Intelligence speaks of what it means to *study* intelligence" (my emphasis). No wonder those of us outside the field have trouble figuring out what AI is really about. My impression is that AI researchers try to study intelligence by building artifacts that will make a convincing show of intelligent behavior. This might be why books on AI methods are all about sophisticated representations and fancy program structures - they're techniques of building more complex (hopefully more intelligent) programs. But this is nearsighted. Intelligence is the *difference* between unintelligent and intelligent behavior. The study of intelligence begins when the programming stops. And on what to do then, the AI textbooks are silent. Now I don't want to spend time talking about the consequences of this failure, Don did that much better than I can. (However, I can't resist throwing in my excuse: programming is fun; science is hard, often boring, work. Science is far more rewarding, though.) What I'm going to discuss in the rest of this note stems from his remark that AI workers are "woefully untutored in scientific method". Assuming for the purposes of discussion that we know enough about intelligence to make principled distinctions between it and stupidity (counterintelligence?), what would the scientific study of intelligence look like? One way of answering this question is to look at some of the enterprises that claim to be scientific, but aren't. The main distinction in the list below is between those fields that are unarguably sciences, and those that fail to be scientific in one way or another. True science, the authentic, natural sciences, are ones like astronomy, geology, biology, physics, or chemistry. False sciences are harder to characterize, but here goes: Here's a list of examples of different claimants to the name "science"; mostly impostors, all of them can be called "quasi-sciences". By looking at them, we can gain some sense of what qualities are necessary for real sciences, since the quasi-sciences don't have them. * Fraudulent sciences: Creation Science, Lysenkoism, Scientology (the most generous thing I can say about these is that they appear to proceed by trusting exceptional, one-of-a-kind reports, and denying persistent, repeated, quantitative, skeptical observations. In rhetoric this is called "appeal to authority.") * Trivial sciences: Clairol Science, barbeque science, accelerator science (Clairol Science has discovered a new way to make your hair silkier and more full-bodied. Barbeque science has conclusively determined that mesquite smoke is superior to hickory smoke. We need to build the superconducting supercollider so America won't fall behind in accelerator science.) * Semi-sciences: Theoretical Physics, Descriptive Linguistics (complementary halves of their respective fields.) * Interdisciplinary Sciences: Materials Science, Neuroscience (characterized by their subject matter not yielding coherently to any single experimental technique or theoretical paradigm.) * Artifact Sciences: Economics, Political Science, Anthropology (Herbert Simon's "sciences of the artificial" - these study artifacts of human society - without civilization, they wouldn't exist. However, civilization is big and complex enough that techniques developed to deal with natural phenomena give useful insights.) * Synthetic Sciences: Mathematics, Computer Science (These study the consequences of small sets of fundamental concepts. Mathematics under Russell&Whitehead and Bourbaki has been "nothing but" an incredibly vast and elegant elaboration of set theory, while [I claim with a certain trepidation] that the fundamental basis of the scientific part of computer science lies in the elaboration of the consequences of the notion of an algorithm.) The authentic, natural sciences, on the other hand, are the body of analytic, experimental studies of phenomena that go on whether or not the experimenter is there to observe them, [philosophers can complain about "naive realism" -- I'll confess to the realism, but not not the naivete] and the results, conclusions, and theoretical relations that tie the studies together. The key concepts here are "experimental" and "objective". If a researcher (or a team of them) isn't doing experiments on some external phenomenon, then it ain't real science. What do you get from real science? Reality. Not wishful thinking, not hallucinations, not mythology, not common sense. (Strictly speaking, what you get is the most compact model of reality consistent with the most reliable, most detailed, widest ranging set of observations.) Uncommon sense. What you don't get is completeness, or even closure. First of all, there's too much knowledge, as anyone with a Ph.D. in a natural science will tell you. Second of all, the universe isn't closed under observation: there's always more detail to examined, further frontiers to be explored, greater complexities to be explained. And most exciting of all, there's the possibility of revolution - that a new model will explain more data, resolve old inconsistencies, or be statable more succinctly, hopefully all at once. The natural sciences generate an interconnected web of explanations that should contain a place for AI, if AI is a science. It's in this explanatory web that people claim to see the bugaboo of reductionism (without which no discussion of scientific method would be complete). Stripped of the argumentative mumbo-jumbo that keeps philosophers in business, a reductionist would claim that a pile of parts on the floor is equivalent to an assembled machine, while a holist would claim that the parts are irrelevant to any description of the machine. Both views are incomplete, but there is indeed an ordering by "is explained in terms of" that reductionists have grabbed onto. Because it's only a partial ordering, I'd like to borrow a term from evolutionary biology and suggest that scientific knowledge has the same kind of familial, clade structure as do charts of the genetic relations among organisms. Reading "<--" as "is used to explain", we have One path through a Cladistic epistemology: Particle Physics <-- Condensed-matter physics <-- Quantum Chemistry <-- Organic Chemistry <-- Molecular Biology/Genetics <-- Developmental Biology <-- Neuroscience <-- Ethology <-- Psychology <-- Cognitive Science <-- Mathematics I would put intelligence in at the same level as mathematics. Congratulations! Scientific AI would be among the most complex of sciences. However, in reality the picture isn't this clean. Aside from those sciences that aren't in a direct explanatory line to intelligence, there are shortcuts among levels due to the logic of experimental science, that makes it possible to do things like manipulate genetic structure and get a behavioral result. But this note is already too long to go into this further, and I've barely alluded to the formal role of the hypothesis. Hope this helps, - George McKee College of Computer Science Northeastern University, Boston 02115 CSnet: mckee@Corwin.CCS.Northeastern.EDU Phone: (617) 437-5204 Usenet: in New England, it's not unusual to have to say "can't get there from here." ------------------------------ End of AIList Digest ******************** 26-Jul-87 23:49:11-PDT,23781;000000000000 Mail-From: LAWS created at 26-Jul-87 23:44:52 Date: Sun 26 Jul 1987 23:42-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V5 #189 To: AIList@STRIPE.SRI.COM AIList Digest Monday, 27 Jul 1987 Volume 5 : Issue 189 Today's Topics: Bibliography - Leff File a55AB ---------------------------------------------------------------------- Date: Sat, 18 Jul 1987 10:37 CST From: Leff (Southern Methodist University) Subject: Bibliography - Leff File a55AB %A Rudolph E. Seviora %T Knowledge-Based Program Debugging Systems %J IEEE Software %V 4 %N 3 %D May 1987 %P 20-32 %K AA08 %X Divides debugging, whether done by human or by computer into three categories: a) those that look at the code and compare to specifications b) those that look at the output c) those that look at the internal trace In the latter category, there exists only one system, MTA which is a PROLOG based system to view internal traces from communication based software which is often done using finite-state machines. Falosy is an example of a system that tries to debug by comparing the output and then reasoning to the program. Proust, Laura, and Pudsy are example of systems that look at the code and compare it to the specification. %T New Products %J ComputerWorld %D APR 13, 1987 %V 21 %N 15 %P 44 %K H01 AT02 AI06 %X AST's new read right software associated with it's page scanner will handle mixed fonts, variable character sizes and spacing and reduced and enlarged photocopies. %T Breaking the Lisp Language Barrier with COBOL %J Mini-Micro Systems %D May 1987 %P 27 %V 20 %N 5 %K AA06 AT02 %X Cullinet has announced a series of expert system programs based on COBOL, OrderEXL, SalesEXL, VoiceEXL and DMS applications expert. %A John Goach %T Coming: A System for Real Time Dialog %J Electronics %D APR 30, 1987 %P 38-39 %K AI05 %V 60 %N 9 %X Spicos, handles 1000 word vocabulary, in continuous speech. The system gives spoken answers and also uses the context of the word to help identify it. %A Henry Eric Firdman %T Which Comes First -- Development Or Specs %J ComputerWorld %D APR 13, 1987 %V 21 %N 15 %P 69-73 %K O02 AI01 rapid prototyping AA08 %X Discusses whether a prototype model of development is appropriate for expert systems as well as software projects in general and what actions are appropriate and not appropriate under such a model of development. %A Jean S. Bozman %T MDBS Develops Guru for VAX %J ComputerWorld %V 21 %D May 25, 1987 %N 21 %K H01 T03 AI09 AT02 %X GURU, which currently runs on PC's, will be ported to both VMS and Ultrix with costs of $17,000 to $60,000 %A Charles Babcock %T Quick and Dirty Fixes May Work Best %J ComputerWorld %V 21 %D May 25, 1987 %N 21 %P 25 %K AA08 %X A study has shown that advanced development techniques leads to more costly maintenance, not less. "Programmers who perform impromptu fixes without checking documentation may be just as effective as those who follow more structured approaches." They also found that a greater percentage of the requests made by users were implemented in systems where the user under the system. %A Rosemary Hamilton %T DG Courts LISP Machine %J ComputerWorld %D APR 13, 1987 %V 21 %N 15 %P 93+ %K H02 AT16 Lisp Machine Inc. LMI Data Bankruptcy General %X Data General made an offer to buy Lisp Machine, which is subject to approval of an LMI creditor's committee. Lisp Machine has filed under Chapter 11 of the U. S. Bankruptcy Law %A Paul Wallich %T Putting Speech Recognizers to Work %J IEEE Spectrum %D APR 1987 %P 55-57 %K AI05 H01 %V 24 %N 4 %X List of current products available. .DS L SSB-1000, Speaker dependent isolated word, 144 words, 95% accuracy, $250 VoDialer, Speaker Dependent isolated word, 48 words, 95% accuracy, $349 (for allowing cellular telephone users to dial numbers) Dragon Systems Voice Scribe, Speaker-dependent, isolated word, 1000 words, $1195 IBM, Speaker-dependent isolated word, 64 words, 95-98% accuracy, $1195 Intel, Speaker Dependent Isolated Word, 200 words Interstate Voice Products, Speaker-dependent, connected speech, 400 words, 98% accuracy, $395 Interstate Voice Products, Speaker-dependent continuous speech, 100 words, 99% accuracy, $4000 Kurzweil Applied Intelligence, Speaker-dependent, isolated word, 1000 words, $6000 NEC, SAR 10, Speaker-dependent, isolated word, 250words , 98% accuracy, $599 NEC, SR10, Speaker-Dependent, isolated words, 128 words, 98% accuracy, $600 NEC, DP-200, Speaker-Dependent, connected speech, 150 words, $7500 Speech Systems, speaker-dependent, connected speech, 20000 words, 90% accuracy, $5000.00 TI, speaker-dependent, isolated word, 1000 words, $995.00 Voice Connection, speaker-dependnet, isolated-word, 400 words, 98% accuracy, $495.00 Voice Control Systems, Speaker-independent, isolated word, 40 words, 98.5% accuracy, $1000.00 Votan, Speaker-independent isolated-word, 13 words, 98% accuracy, $1350 Votan, speaker-dependnet, continuous speech, 640 words, 94 percent, $1200.00 .DE %A H. Sardar Amin Saleh %T Artificial Intelligence and Computer Aided Design in Civil Engineering %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 781-789 %K AA05 AI01 T02 %A Shuichi Fukuda %T Development of an Expert System for the Design Support of an Oil Storage Tank %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 791-796 %K AA05 AI01 GA01 %X This system copes with such as issues, corrosions, local regulations and interfaces with numerical software to assist in the design of oil storage tanks. %A John F. Brotchie %A Ron Sharpe %A Bertil Marksjo %A Michael Georgeff %T Introducing Intelligence and Knowledge Into CAD %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 797-810 %K AA05 optimization quadratic programming %X discusses applications of AI to quadratic programming with nonconvex solutions. %A U. Flemming %A R. Coyne %A T. Glavin %A M. Rychener %T A Generative Expert System for the Design of Building Layouts %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 811-821 %K AA05 kitchen bathroom generate and test AI09 DENDRAL %X This system designs kitchen and bathrooms using an approach based upon DENDRAL with a generator generating possible layouts and a tester evaluating them against the constraints. %A S. F. Jozwiak %T Applications of Artificial Intelligence in Structural Optimization %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 823-834 %K AA05 AI04 %X AI techniques are used to reduce the computer time in determining the optimal positions of the nodes in a three dimensional truss. This type of optimization is done by setting up a constraint corresponding to each member of the truss insuring that that member does not bear unacceptable stresses. This work compares known truss structures against the one being optimized to determine which elements are likely to have stresses lower than adjacent stresses so they don't need to have their stresses computed. Same content as a paper appearing in \fIComputers in Structures\fR by the same author. %A T. J. Ross %A F. S. Wong %T Structural Damage Assessment Using AI Techniques %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 835-846 %K AI01 AA05 AA18 %X This system helps assess the possible damage to buried concrete boxes from nearby nuclear explosions %A Peter W. Mullarkey %T A Geotechnical KBS Using Fuzzy Logic %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 843-860 %K AA05 AI01 O04 %X This system helps interpret the results of cone penetrometer tests in determining the soil conditions where some structure will have its foundation. %A Kenneth R. Maser %T Automated Interpretation of Sensor Data for Evaluating In-Situ Conditions %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 861-888 %K AA05 radar signal interpretation AI06 AI01 %X The system helps model bridge deck deterioration using ground penetrometer studies. The expert system deals with considerations from radar signal analysis, radar/concrete physics, and bridge engineering. %A Yoon-Pin Foo %A Hideaki Kobayashi %T A Framework for Managing VLSI CAD Data %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 889-898 %K AA05 AA09 AA16 %X compares a frame based system using is-a type inheritance with INGRES database approach showing that operations are performed about sixty percent faster in their frame based system. %A Nikhil Balram %A William P. Birmingham %A Sean Brady %A Robert Tremain %A Daniel P. Siewiorek %T The MICON System for Single Board Computer Design %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 899-910 %K AA04 AI01 %X This system constructions microcomputer boards. It deals with the problems of interfacing IO chips from one family such as a Z80 SIO chip to some other microprocessor. The system also handles analog type constraints such as bus propagation, etc. %A Jeffrey L. Dawson %T Excirsize - An Expert System for VLSI Transistor Sizing %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 911-916 %K AA04 AI01 %X Describes how to size transistors for NMOS fabrication where the goal is to achieve some propagation constraint at minimum power consumption. %A Ravi Malhotra %A Ken Chao %A Osama Mowafi %T A Knowledge-Based System for Network Communication Design %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 917-924 %K AA08 %X discusses applications to designing the backbone part of a network consisting of high speed communication paths and the access part consisting of low speed lines connecting to various cities. %A Stuart C. Shapiro %A Sargur N. Srihari %A Ming-Ruey Taie %A James Geller %T VMES: A Network-Based Versatile Maintenance System %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 925-936 %K AA21 %X Shows how in a structure based diagnostic expert system, to save memory for common parts. I. E., if there are several op-amps in the circuit, one doesn't want to store the capacitors, etc. for each op-amp in the memory that comprise the op-amps. Techniques are developed to phase in the detailed description of a part when needed to save computer time. The article also discussed the graphic interface to the VMES and shows how the system chooses what to display and how to arrange these items on the screen. %A Tao Li %T Heuristic Search in Digital System Diagnosis %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 937-946 %K AI03 AA04 %X shows a variation of the A* algorithm for use in detecting faults in sequential circuits. The article also shows how to handle circuits that are not "resettable," there is no signal that is guaranteed to force the system into a known state. Various theoretical results regarding such fault detections are also provided. %A William P. C. Ho %T A Plan Patching Approach to Switchbox Routing %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 947-958 %K AA04 AI09 %X When a conventional routing system reports failure, i. e. all the rules that it has available have been tried, this system will come in and try and patch the almost completed solution into a successful routing of the switchbox. %A Bryant W. York %T KBTA: An Expert Aid for Chip Test %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 959-970 %A Jozsef Vancza %T CODEX: A Coding Expert for Programmable Logic Controllers %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 971-984 %K AA05 %X This system takes a generic description of the control task to be performed and translates it into the language for a specific make and model of programmable controller. %A Ernesto Guerrieri %A Vinod Grover %T Octtree Solid Modeling with Prolog %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 985-1002 %K T02 AA05 %X Shows how the OCTTREE data structure for representing objects can be entered as Prolog acts and unions, interference checking and neighbor finding are performed upon them. %A G. Goldbogen %A D. Ferrucci %T Extending the Octree Model to Include Knowledge for Manufacturing %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1003-1012 %K T01 T02 AA26 %X Describes feature extraction algorithms on octtrees for features such as hole boundaries. %A C. B. Bouleeswaran %A H. G. Fischer %T A Knowledge Based Environment for Process Planning %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1013-1028 %K AA26 AI01 %X describes a system that generates process plans for rotational parts such as screws. The system supports integrated design of the part to be machined and the manufacturing process to use on it. %A Joao P. Martins %A Stuart C. Shapiro %T Hypothetical Reasoning %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1029-1042 %K AI16 %X Discusses a generic purpose tool to allow users to raise hypotheses, reason from them, discard various hypotheses and perform the appropriate truth maintenance. The system uses contexts to avoid backtracking. %A Robert Milne %T Fault Diagnosis Using Structure and Function %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1043-1054 %K AA19 AI01 %X A troubleshooting paradigm called the "Theory of Responsibilities" is introduced and applied to testing circuits. It works from "second principles" in assigning responsability for various parts of the output waveform to various components of the circuit. %A D. Sharma %A B. Chandrasekaran %A D. Miller %T Dynamic Procedure Synthesis, Execution, and Failure Recovery %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1055-1072 %K AA05 nuclear power plant AI01 AI09 %X Describes a system for planning failure recovery, synthesis, monitoring for nuclear power plants. A comparison of the "event-oriented" and "function oriented" approaches to nuclear power plant management is provided. The nuclear industry is shifting to the latter in reaction to the TMI difficulties. The implications of this for expert system applications and an example from reactor scram concerns are also provided. Various plan templates and blackboards are used in processing. The final expert system consists of system specialists, specialists in various kind of undesirable events and specialists in various kind of goals such as reducing radioactivity. %A B. Demo %A M. Tilli %T Expert System Functionalities for Database Design Tools %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1073-1082 %K AI01 AA09 %X Discusses the CARS system, an expert system for the design of databases. %A Geoffrey D. Gosling %A Anna M. Okseniuk %T SLICE - A System for Simulation Through a Set of Cooperating Experts %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1083-1096 %K This paper describes simulation tools to investigate the application of expert systems to aircraft control environments. %A T. J. Grant %T Maintenance Engineering Management Applications of Artificial Intelligence %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1097-1122 %K AA21 AI01 AA05 %X This is a survey of potential applications of artificial intelligence to managing the maintenance of aircraft. An interesting comment is that twenty percent of all faults are novel (noone ever saw them before). These faults required twice as many repair hours to fix as the average fault. For any given diagnostician, the number of faults that he never saw before approaches sixty percent. It is interesting to note that 63 percent of the Royal Air Force's manpower is employed doing maintenance. %A Benoit Faller %T Expert Systems in Meteorology %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1123-1127 %K AA16 AI01 %X presents expert systems for forecasting airport fog-in conditions, storm forecasting and avalanche risks. Fogs are predicted in the afternoon for the following morning. %A Karl-Erik Arzen %T Expert Systems for Process Control %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1127-1138 %K AA05 AI01 %X Uses of expert systems with various control systems concepts such as the Ziegler-Nichols auto-tuner, smart PID controller and the Nichols auto-tuner. %A Atsumi Imamiya %A Akoio Kondoh %A Akiyoshi Miyatake %T An Artificial Intelligence Approach to the Modeling of the User-Computer Communications %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1139-1151 %K AA08 AA15 AI01 %X describes a system to automate the production of help systems for software. %A John R. Hogley %A Alan R. Korncoff %T Artificial Intelligence in Engineering: A Revolutionary Change %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1155-1160 %K AA05 AI01 %X This is a general article devoid of technical content. %A Kai-li Kan %T Expert Systems in Telecommunications Network Planning and Design %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1161-1165 %K AI01 AA08 %X discusses the implications of expert system in design networks. The gentleman belongs to the "Strategic Technology Assessment" department of Pacific Bell. %A Ye-Sho Chen %T Expert System for On-Line Quality Control %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1165-1174 %K AI01 AA05 AA21 automotive O04 Pareto %X Discusses a diagnostic system for automobile brakes. The system uses Pareto optimality to assist in uncertainty calculus. %A K. M. Chalfan %T An Expert Executive Which Integrates Heterogenous Computational Programs %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1175-1174 %K AI01 AA05 aerospace preliminary design %X Discusses a proposed system to automate weight, aerodynamics, propulsion and performance codes in the preliminary design of airplanes. %A A. Kissil %A A. Kamel %T An Expert System Finite Element Modeler %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1179-1186 %K AI01 AA05 %X Discusses an expert system for use in generating meshes for finite elements. Includes a discussion of heuristics that will generate a mesh with a desired accuracy. This is done by comparing a parametric distortion of a region whose stresses are known with the unknown region. %A Paul F. Monaghan %A James G. Doheny %T Knowledge Representation in the Conceptual Design Process for Building Energy Systems %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1187-1192 %K HVAC AA05 AI01 %X discusses using expert systems and hierarchies in the design of HVAC systems (Heating, Ventilation and Air Conditioning) %A Prem Kumar Kalra %T Development of Expert System for Fault Diagnosis in HVDC Systems Using Spectral Approach %B Applications of Artificial Intelligence in Engineering Problems %E D. Sriram %E R. Adey %V 2 %I Computational Mechanics Publications %C Woburn, Massachussetts %D 1986 %P 1193-1198 %K AA21 AA05 AI01 %X discusses using Fast Fourier Transform, Fast Walsh Transform and expert systems to help diagnose high voltage DC and analog systems ------------------------------ End of AIList Digest ******************** 28-Jul-87 23:34:14-PDT,14082;000000000000 Mail-From: LAWS created at 28-Jul-87 23:25:15 Date: Tue 28 Jul 1987 23:21-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList V5 #190 - Msc. To: AIList@STRIPE.SRI.COM AIList Digest Wednesday, 29 Jul 1987 Volume 5 : Issue 190 Today's Topics: Queries - AI/Graphics & Examples of KEE Frames & Problem Recognition in Prolog Databases & NLP Front Ends to INGRES, Policy - Virtual Sublists, Philosophy - Natural Kinds ---------------------------------------------------------------------- Date: Mon, 27 Jul 87 14:03:36 BST From: mcvax!ux.cs.man.ac.uk!arnold@seismo.CSS.GOV Reply-to: thoward@uk.ac.man.cs.cgu Subject: AI/Graphics: help wanted I am currently investigating what work has been done on connecting/ integrating AI methods and computer graphics. I would be very grateful if anyone can send me any references, or bibliographies (or comments!) etc in this area. If there's enough interest, I will summarise responses. Please mail to me directly, *not* to the source of this posting, as it's not my own account. Thanks... Toby Howard Janet: thoward@uk.ac.man.cs.cgu University of Manchester ARPA: thoward%cgu.cs.man.ac.uk@cs.ucl.ac.uk Computer Graphics Unit Phone: +44 61 273 7121 x5429/5406 ------------------------------ Date: 28 Jul 87 04:46:37 GMT From: munnari!uqcspe.OZ!twine@uunet.UU.NET (Steven Twine) Subject: Examples of KEE frames Requested I am currently revising a semantic analysis of KEE's frame language. By semantic analysis, I mean trying to answer the question What facts does X encode about the current Universe of Discourse where X is each of the syntactic ingredients in a KEE knowledge base (units, slots, links etc). This is not as simple as it seems, because a given KEE construct can represent many different things (as Brachman showed for IsA links). Anyway, in revising this paper, I would like to add many more examples of KEE structures that have been used in practice, for the purpose of analysing the facts that they encode. I am particularly interested in any ambiguous or otherwise tricky examples that I can test my interpretations out on. I would appreciate any examples of KEE units etc that people could send me for this purpose (examples in other frame languages may also be useful, but KEE is preferred) All senders will get a lovely acknowledgement at the end of the paper (what an incentive!) as well as my heartfelt gratitude. Thanks in advance, folks! ========================================================================= Steven Twine, ARPA: twine%uqcspe.oz@seismo.css.gov Department of Computer Science, ACSnet: twine@uqcspe.oz University of Queensland, UUCP: seismo!munnari!uqcspe.oz!twine St Lucia, 4067. CSNET: twine@uqcspe.oz AUSTRALIA. JANET: uqcspe.oz!twine@ukc ------------------------------ Date: 26 Jul 87 20:37:16 GMT From: dartvax!balu.UUCP@seismo.css.gov (Balu Raman) Subject: Problem recognition in Prolog database I am working on recognizing problem instances in Prolog database. The problems can be typical Graph-color, Linear Programming Problem, Critical Path Problems etc.etc. Does anybody in the netland have references, pointers ,prolog programs to do what I am trying to do. thanks in advance. Balu Raman. ------------------------------ Date: Mon, 27 Jul 87 08:37:24 PDT From: vor!cris%esosun.UUCP@sdcsvax.ucsd.edu (Cris Kobryn) Subject: NLP Front-Ends to INGRES I am interested in developing an NLP front-end to INGRES. Lest I reinvent: Is there any "stock" software which already does this? (INTELLECT does not *currently* accommodate INGRES; I've heard "DataTalker" mentioned as a possibility, but have no details--capabilities, company name, phone#, etc.) Re building an NLP front-end: Prolog's DCG's (Definite Clause Grammars) seem to provide an attractive tool to construct an NLP front-end. I would appreciate feedback re their effectiveness, and pointers to work done or being done relevant to this interest. I will be glad to summarize and post if the response merits it. -- Cris Kobryn +----------------------------------------------------------------------------+ | Cris Kobryn UUCP: {sdcsvax|seismo}!esosun!cris | | Geophysics Division, MS/22 ARPA: esosun!cris@seismo.css.gov | | SAIC SOUND: (619)458-2697 | | 10210 Campus Point Drive | | San Diego, CA 92121 | +----------------------------------------------------------------------------+ ------------------------------ Date: 27 Jul 87 14:28 PDT From: Ghenis.pasa@Xerox.COM Subject: PROPOSAL: We need "virtual sublists" The recent meta-discussion on what to include in the Digest was rather similar to the one about whether to include the AI Expert code listings. At that time I made a proposal that may have drowned in the noise. I still think it would solve the filtering problem so here it goes: PROBLEM: You can't tell what is inside the digest until you start reading it. The title is non-descriptive. How does an AIList reader filter unwanted topics? If a reader has an unsophisticated mail reading channel, there is an irritating time cost to opening an unwanted 20,000 character message. This is even worse for folks who read their mail through a modem connection. Proposing the creation of a new list for each topic that generates a large mail volume is not only unrealistic but also unnecessary. SOLUTION: The moderator is already thoughtful enough to segregate topics so that each digest is fairly homogeneous. Now if only the "Subject:" line could read AIList V5 #183 - Symbol Grounding instead of AIList Digest V5 #183 then it would be easy to filter topics even with the crudest of mail programs, and our personal archives would also be much more descriptive at the table-of-contents level. I believe that this scheme would address the objections of folks who voted against continuing to distribute symbol grounding messages or source code listings. MODERATOR: Would this be a difficult change to implement? FELLOW READERS: Is this proposal missing the point? Is there anything else we could do to better prepare for the next large discussion? Should we move this discussion to the META-META-DISCUSSIONS list? :-) Pablo Ghenis Xerox Artificial Intelligence Systems Educational Services [This has been suggested several times, by several people, so I might as well give it a try. I am reminded, though, of a parody of Reader's Digest that condensed an entire Hemmingway novel to the word "Bang!". A good many digests will have to be tagged as "Msc.", including this one. I really don't see the advantage in the longer subject line, but perhaps that is because my mailer clips the subject at about 40 characters. The cost of examining the full Topics section is only about one page of data. (Are there really mailers out there that let you read the subject line without the cost of "pulling in" the entire digest?) What is really needed here is an intelligent mail-reading system. I'm sure that special digest-reading commands could -- but probably won't -- be added to any of our mailers. Even better would be an intelligent Information Lens system. Won't someone take this on as an AI project? -- KIL] ------------------------------ Date: 27 Jul 87 09:45:19 edt From: Walter Hamscher Subject: Natural Kinds (Re: AIList Digest V5 #186) Your functional description of "chair" does capture more of "what's essential to chairs" than the structural description could. Some quibbles, however. First, it includes couches since it doesn't say that it's for exactly one person. Second, it doesn't seem to include "Balenz" chairs, those kind in which the person rests on his/her shins, since the "support for one's back" is rather indirect -- what they do is to make it easier to balance the spine by tilting the pelvis forward. Third, some people might say that Balenz chairs aren't chairs at all, but stools, because the back support is indirect -- the point being that the functional description might have to take into account who's saying what about chairs to whom. Probably, other Ailist readers will come up with more borderline cases, which brings me to the speculation that functional descriptions may end up with as many exceptions as structural descriptions do. ------------------------------ Date: Mon, 27 Jul 1987 11:16 EDT From: MINSKY%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: Natural Kinds (Re: AIList Digest V5 #186) I agree: 1. Yes, I think we'd all agree that a chair is for 1 person to sit on. 2. The boundary is fuzzy, indeed, and some people might not consider a Balenz chair to be a chair. 3. Yes, indeed, the "functional description" does indeed depend on whose "intention" is ivolved, and upon who is saying what to whom. My point is not that such terms can be defined in foolproof, clear-cut ways. There are really two sorts of points. 1. You can get much further in making good definitions by squeezing in from both structural and function directions - and surely others as well. 2. In Society of Mind, section 30.1 I discuss how meanings must depend on speakers, etc. As Ken Laws remarked, we should not be too hasty to thank philosophers for concept of "natural kind". McCarthy make useful remarks about penguins, which form a clear-cut cluster because of the speciation mechanism of sexual reproduction. The class is un-fuzzy even though, as McCarthy notes, penguins have properties that scientists have not yet discovered. But then, I think, McCarthy defeats this clarity by proceeding to discuss how children learn about chairs - and tries to subsume this, too, into natural kinds. He describes what seems clearly to be not "natural" aspects of chairs, but the clustering and debugging processes a child might use. My conclusion - and, I'd bet, Ken Laws would agree - is that the concept of "natural kind" has an illusory generality. It seems to me that, rather than good philosophy, it is merely low-grade science contaminated by naive, traditional common sense concepts. The clusters that have good boundaries, in the world, usually have them for good - but highly varied reasons. Animals form good clusters because of Darwinian speciation of various sorts. Certain metals, like Gold, have "natural" boundaries because of the Pauli exclusion principle which causes things like periodic tables of elements. Philosophers like to speak about gold - but their arguments won't work so well for Steel, whose boundary is fuzzy because there are so many ways to strengthen iron. All in all, the clusters we perceive that have sharp boundaries are quite important, pragmatically, but exist for such a disorderly congeries of reasons that I consider the philosophical discussion of them to be virtually useless in this sense: the class of clusters with "suitable sharp boundaries" to desaerve the title "natural kinds" is itself too fuzzy a concept to help us clarify the nature of how we think about things. ------------------------------ Date: Mon, 27 Jul 87 09:57:26 MDT From: shebs@cs.utah.edu (Stanley Shebs) Reply-to: cs.utah.edu!shebs@cs.utah.edu (Stanley Shebs) Subject: Re: Natural Kinds (Re: AIList Digest V5 #186) In article MINSKY@OZ.AI.MIT.EDU writes: >About natural kinds. In "The Society of Mind", pp123-129, I propose a >way to deal with Wittgenstein's problem of defining terms like "game"- >or "chair". The basic idea was to probe further into what >Wittgenstein was trying to do when he talked about "family >resemblances" and tried to describe a game in terms of properties, the >way one might treat members of a human family: build, features, colour >of eyes, gait, temperament, etc. >[... details of Wittgenstein vs Minsky :-) ...] >I would appreciate comments, because I think this may be an important >theory, and no one seems to have noticed it. [...] I recently finished reading "Society of Mind", and quite enjoyed it. There are a lot of interesting ideas. There are also many that are familiar to people in the field, but with new syntheses that make the ideas much more plausible than in the past. I had been getting cynical about AI, but after reading this, I wanted to go and hack out programs to test the hypotheses about action, and memory, and language. But there's a serious problem; how *can* these hypotheses be tested? The society of mind follows human thinking so closely that any implementation is going to be a model of human minds rather than minds in general, and will probably be handicapped by being too small and simple to be recognizably human-like in its behavior. Tracing a mind society's behavior will generate lots of data but little insight. So my ardor has been replaced by odd moments speculating on tricky but believable tests, and a greater appreciation for people interested in a more formal approach to minds. Getting down to specifics, the theory about recognition of objects by either structure or functions was one of the parts I really liked. A robot should be able to sit on a desk without getting neurotic, or to sit carefully on a chair that's missing one leg... stan shebs ------------------------------ End of AIList Digest ******************** 29-Jul-87 22:03:07-PDT,12036;000000000000 Mail-From: LAWS created at 29-Jul-87 21:56:03 Date: Wed 29 Jul 1987 21:51-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList V5 #191 - LISP Techniques To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 30 Jul 1987 Volume 5 : Issue 191 Today's Topics: Techniques - Graphics-AI References & Garbage Collection Suppression ---------------------------------------------------------------------- Date: Tue, 28 Jul 87 13:56:14 pdt From: Eugene Miya N. Subject: Re: Graphics-AI bibliography >I am currently investigating what work has been done on connecting/integrating >AI methods and computer graphics. I would be very grateful if anyone can >send me any references, or bibliographies (or comments!) etc in this area. > - Toby Howard - Computer graphics: get the current edition of ACM Computer Graphics (the Quarterly). It has the yearly bibliography in computer graphics (June 1987). There is a biblio for June 1986 which over the year 1985. Recently, we had a meeting where we had a speaker from SRI cover some of the common ground (Sandy Pentland) because we perceived that the AI people were reinventing what the graphics people invented 20 years ago. --eugene miya Bay Area ACM/SIGGRAPH ------------------------------ Date: Wed, 29 Jul 87 11:58 PDT From: nesliwa%telemail@ames.arpa (NANCY E. SLIWA) Subject: Garbage Collection Suppression (Response Summary) My thanks to all the respondants to my question about garbage collection suppression. As several people asked for the results, I'm posting it for all: Date: Friday, 17 July 1987 07:32-CDT From: nancy at grasp.cis.upenn.edu (Nancy Orlando) Are there any "accepted" methods of writing code that minimize a LISP's tendancy to garbage-collect? I don't mean a switch to turn it off; just a means of minimizing the need for it. I'm dealing particularly with DEC VAX lisp. I have assumed that iteration as opposed to recursion was one way; is this correct? From: Chester@UDEL.EDU Subject: Re: Garbage Collection Suppression The direct way to avoid garbage collection in lisp is to define your own `cons function that prefers to get cell pairs from an `available list', calling the regular `cons' only when the `available list' is empty. A `reclaim' function that puts cell pairs on the `available list' (using `rplacd') will be needed also. See any book on data structures. The technique can be used for cell pairs and gensym atoms, if needed, but in my experience, not with strings or numbers. String manipulations can usually be avoided, but a program that crunches a lot of numbers cannot avoid consuming memory and eventually triggering garbage collection (at least in VAX lisp). I wish there were some way for a user to reclaim numbers so that they could be reused as cell pairs can. If so, I could write all my lisp programs so that they don't need to garbage collect. It would also be nice to have a built-in `reclaim' function that would work in conjunction with the built-in `cons'; it would be dangerous for novices, but handy for the experienced. By the way, recursion in itself doesn't cause garbage collection; VAX lisp is smart enough to reclaim the memory used for the function stack automatically. Daniel Chester chester@dewey.udel.edu Date: Mon, 20 Jul 87 01:36:44 PDT From: woutput@ji.Berkeley.EDU (Andrew Purshottam) Subject: Re: Garbage Collection Suppression Forgive me if my response is too trivial, but you ommited the most important technqiue for reducing gc use, limiting the use, implict and explict, of cons. Particularly nasty is the use of append or append1 (not sure what that is called in CL) to build up a list by adding elements to its end. This method uses O(n^2) cons cells, where n is the length of list built. Standard solutions include the use of "accumulators", arguments which hold a partial result which is modified in inner recursions and finally returned as value when the function returns; building the list backword and maybe reversing it at end; nconc, which uses O(n^2) time but only O(n) space; or tconc structures, which keep a pointer to the end of the list. (In prolog we have a cute method avail, putting an uninstantiated element at the end of the list, effectively a "hole" that can be filled by an element and another hole). Not also that some popular functional programming techniques, particularly those involving streams and higher order procedures are quite greedy in cons cells, as they build intermediate lists, most of whose elements are thrown away. The apply-append-mapcar trick, the set functions like (filter 'pred 'list), union, and intersect all do this if implemented in the obvious way, with the sets represented and fully computed lists. The Black Book (Charniak/McDerrmot, AI Programming) discusses more eff. ways to deal with this using generators, where no more elements are computed than needed. (See also Abelson/Sussman for a very readable (we inflict it on freshman!) discussion of delay and force). Again, excuse if this is too simple, no offense intended. Andy -- Cheers, Andy (...!ucbvax!woutput woutput@ji.berkeley.edu) (cond ((lovep you (quote LISP)) (honk)) (t (return ()))) Date: Mon, 20 Jul 1987 05:32 CDT From: AI.DUFFY@R20.UTEXAS.EDU Subject: Garbage Collection Suppression No. You make garbage when you create data structures. Recursion v. iteration has nothing to do with it, unless VAXlisp is more brain-damaged than I already know it to be. Are there other techniques? Use destructive list operations (e.g., NCONC instead of APPEND) when you can. If you have any arrays, structures, etc., that you are using temporarily, you can resource them (make a bunch of them and push them onto a list, pop one off when you want to use it, and when you are finished with it, nullify its slots and push it back onto the list). Your best bet, of course, is to get more memory. Date: 20 Jul 87 09:32:01 edt From: Walter Hamscher Subject: Garbage Collection Suppression Iteration vs. Recursion is orthogonal. By using recursion where you could have used iteration, you may be using _stack_ space, but that's trivially `garbage collected' every time you return from a function (this is non-tail recursion I'm talking about). The only surefire way to reduce garbage collection is to call CONS and and MAKE-ARRAY (and things that call them) less often. There are a number of implications of that for coding style (e.g. pass functions down instead of passing consed structures up), but using iteration is not one of them. Date: Mon, 20 Jul 1987 10:12 EDT From: "Scott E. Fahlman" Subject: Consing Nancy, In order to avoid GC's, you have to write your code in a way that avoids consing up data structures, especially in inner loops. In the Vax Common Lisp, I think that recursion is faster than the equivalent iteration, but consing is not the reason; what you're seeing is the difference between access to just a few variables (in registers or in the cache) versus spreading out copies of those registers on the stack, with all the associated memory references for pushing and popping. How to avoid consing in any given Comon Lisp is a complex topic. Perhaps the DEC people have some training materials on how to do this in their Lisp. But there are a few things to watch for: Make sure all your code is compiled. A lot of Lisps cons furiously in the interpreter while consing very much less in compiled code. If you are consing up vectors and strings in some inner loop solely for communication with other routines, consider passing the info in a single pre-allocated vector instead. Some Lisps cons when passing &rest args and &keyword args. Check this. Often a bit of Consing can make code clearer and easier to maintain. Find who is doing the consing that is bothering you and squeeze that part of the code for maximum efficiency; don't just squeeze everything, because maintainability will be harmed. -- Scott Fahlman Date: Mon, 20 Jul 87 10:15:53 EDT From: Mario O. Bourgoin Subject: Re: Garbage Collection Suppression Hello, What you want to do is avoid storage allocation operations. Most of the methods for doing this are implementation dependant. For example, in Scheme iteration constructs expand to tail-recursive calls so there's no point in trying to change function calls to do-loops. Furthermore, good Lisp implementations optimize function calls since they are the most used operation; they are usually cheaper than the looping alternatives. Lisp compilers can usually do excellent optimizations if you use the implemetation's features. For example, ZetaLisp offers a LOOP iteration macro which allows the programmer to communicate to the compiler the necessary information for the latter to produce the most efficient code possible. What you can do reliably is to avoid using operations that cons a lot such as `append' and use their structure modifying alternatives such as `nconc'. You should be careful to write your programs with the modifying operations from the beginning to avoid encountering problems with them if you change over from the consing operations. Remember that operations such as the arithmetic functions must allocate storage for their result. It might be worth your while to code basic operations and inner loops in another language such as `C' to avoid allocation. --Mario O. Bourgoin (This next was is response to a follow-up request of mine, asking if call-outs to non-lisp external routines helped decrease garbage collection.) Date: Thu, 23 Jul 87 9:11:18 EDT From: Chester@UDEL.EDU Subject: Re: garbagecollection We have no experience with calling out to another language just to do number crunching. My guess is that the overhead of switching languages and of communicating between them and lisp would be too much, but that is just a guess. If you find out differently, let me know. Date: 22 Jul 87 14:28:51 GMT From: "J. A. \"Biep\" Durieux" Subject: Re: Garbage Collection Suppression In article <8707202143.aa23792@Dewey.UDEL.EDU> Chester@UDEL.EDU writes: >The direct way to avoid garbage collection in lisp is to define your own `con >function that prefers to get cell pairs from an `available list' (...). Also handy in many cases (small functions like append, alist-functions, subst) is icons: (defun icons (a d cell) (cond ((and (eq (car cell) a) (eq (cdr cell) d)) cell) (t (cons a d)))) In this way whenever it turns out the new cells weren't really needed, the old ones are used again (as in (append x nil)). Be aware, however, that your copy-function may not work any more if it's defined as (subst nil nil x)! -- Biep. (biep@cs.vu.nl via mcvax) ***************************************************************************** I also noticed that the current (Aug.-Sept. 87) issue of LISP Pointers has two good articles about garbage collection: "Overview of Garbage Collection in Symbolic Computing," by Timothy J. McEntree (TI) and "Address/Memory Management For A Gigantic LISP Environment or, GC Considered Harmful," by Jon L. White (MIT). LISP Pointers subscriptions are available from: LISP Pointers Mary S. Van Deusen, Editor IBM Watson Research PO Box 704 Yorktown Heights, NY 10598 ------------------------------ End of AIList Digest ******************** 29-Jul-87 22:05:19-PDT,8479;000000000000 Mail-From: LAWS created at 29-Jul-87 22:01:27 Date: Wed 29 Jul 1987 21:59-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@STRIPE.SRI.COM US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList V5 #192 - Philosophy To: AIList@STRIPE.SRI.COM AIList Digest Thursday, 30 Jul 1987 Volume 5 : Issue 192 Today's Topics: Philosophy - Philosophy-Bashing & AI as a Science & Natural Kinds & Iconic Representation ---------------------------------------------------------------------- Date: Mon, 27 Jul 87 15:40:12 pdt From: ladkin@kestrel.ARPA (Peter Ladkin) Subject: philosophy-bashing i wish contributors to the ailist who indulge in philosophy could refrain from including diffuse comments alluding to the lack of worth of philosophy. philosophers do the same thing, so it's hard to keep track of who's who. peter ladkin ladkin@kestrel.arpa ------------------------------ Date: 29 Jul 87 15:58:46 GMT From: sbrunnoc@hawk.CS.ULowell.Edu (Sean Brunnock) Reply-to: sbrunnoc@hawk.cs.ulowell.edu (S. Brunnock) Subject: Re: Why AI is not a science Gentlemen, please! (my apologies to any women reading this) AI is a very young branch of science. Computer science as a whole is only a little more than 40 years old. How can you compare AI with mathematics or physics which are thousands of years old? Aristotle made some of the first stabs at elemental chemistry and gravitation. From our enlightened viewpoint, can we call him a scientist? Give it time, its too early to tell. S. Brunnock ------------------------------ Date: 29 July 1987, 14:55:35 EDT From: Andrew Taylor Subject: in defence of penguins (natural kinds) Penguins have been the topic of some discussion. I'd like to correct some some misconceptions. Penguins are not one species, currently they are classified into 18 species. Their inability to fly is not a deficiency. Their wings are merely adapted to a more dense medium, water. They are not the only flightless birds there are 40+ species of flightless birds (0.5% of all bird species). It is not certain penguins are birds. In the past it was believed that they were independently descended from the reptiles. It is possible fossils will be found which will cause this belief to rise again. Penguins may form a clear cut group (order) to ornithologists but people less expert could easily classify other birds of similar appearance and habits (e.g auks) into the same group. Unfortunately species are sometimes not clear cut either. When two populations are separated, then it can be difficult to decide whether they are 1 or 2 species. Biologists often merge or split species in new classifications. People living close to nature (e.g Amazon Indians) have "kinds" which mostly correspond to species. Most of us are content with kinds which lump together a number of species on the basis of superficial similarities. These kinds often differ from the classifications biologists make. Andrew Taylor ------------------------------ Date: Wed, 29 Jul 87 08:43:05 -0200 From: Eyal mozes Subject: Re: natural kinds An important theory that has so far not been mentioned in the discussion on "natural kinds" is the Objectivist theory of concepts. In essence, this theory regards universal concepts, such as "chair" or "bird", as the result of a process of "measurement-omission", which mentally integrates objects by omitting the particular measurements of their common characteristics. The theory takes into account the point mentioned in Minsky's recent message about structure and function, and completely solves Wittgenstein's problem. The theory is presented in the book "Introduction to Objectivist Epistemology" by Ayn Rand, and, more recently, in the paper "A theory of abstraction" by David Kelley (Cognition and Brain Theory, vol. 7 no. 3&4, summer/fall 1984, pp. 329-357). Eyal Mozes BITNET: eyal@wisdom CSNET and ARPA: eyal%wisdom.bitnet@wiscvm.wisc.edu UUCP: ...!ihnp4!talcott!WISDOM!eyal ------------------------------ Date: Wed, 29 Jul 87 08:03:56 EDT From: powell%mwcamis@mitre.arpa Subject: Natural Kinds Minsky's notion of natural types involving both structure and function does seem plausible. One could think of each natural type as a bipartite graph where one node class represents structural components and where the other node type represents each function of the natural type. Connections between the two node classes would represent (in a crude way) the way in which portions of each class relate to the nodes of the other class. Even more specifically, the entire design foundations as would be recorded in the data dependency net of an ATMS recording the design process (function to structure) would capture still more about the natural type. This seems like a bizarely specific way to define a hazy notion like natural types, but it does appear to follow naturally from Minsky's proposal. ------------------------------ Date: Wed 29 Jul 87 11:28:26-PDT From: Ken Laws Subject: Structure, Function, and Intention Minsky's initial message described function (of a chair) in terms of intended use. I don't believe he elaborated, but it seems obvious that it could be either the designer of the chair or the user who provides the intention. (For instance, a chair designed for one person does not become a couch just because two kids sit on it at the same time.) Semantic classification thus requires at least three viewpoints: structure, intended function, and perceived or implemented function. -- Ken ------------------------------ Date: Wed, 29 Jul 87 16:11:23 edt From: amsler@flash.bellcore.com (Robert Amsler) Subject: Re: Structural and Functional descriptions Another division of information which I find significant is that of visual vs. the combined structural and functional descriptions. While a visual description might be termed `structural' I think there is a significant difference. Visual information, i.e. information obtained from looking at a visual still or moving image of an object, is often not available in pre-recorded structural form. It `may' be possible to describe visual information in symbolic text, but it can prove very hard to extract it from existing descriptions because there is so much visual information to represent and often the description doesn't contain the key element needed to answer a question. I first encountered this when looking at the information dictionaries present for a word such as `horse'. They give definitions of all the parts of a horse, but you cannot assemble a horse from these part definitions accurately enough to answer a simple question such as whether the horse's head is higher than its tail? (Dictionaries almost universally have an illustration for a horse, which suggests they know something about how hard it is to describe one by definitions only). Initially I saw this as demonstrating the complimentarity of visual and definitional information, much in the same manner that Minsky sees the complimentarity of the structural and functional descriptions. But now, it looks to be a more basic problem. Even if you could assemble a horse from the definition plus the static visual knowledge (e.g. add coordinates and a wire frame model of a horse to the description), I can't animate it well enough to answer questions (Are all the feet ever off the ground simultaneously while running?) This probably suggests a simulation as the correct representation, but often a simulation is really just a means of displaying the visual representation of the object so you can perform the observation needed on the simulated entity rather than on the real entity. What this seems to imply is that ultimately the `description' of an object should be a simulation accurate enough to permit direct observation and generation of the functional and structural information we know about the object? ------------------------------ End of AIList Digest ********************