lots to be changed according to the problem, or do the two views require such fundamentally different implementations that it's just better to stick to one or the other? Is it possible to do work using frames without being concerned about the particular view imposed by the frame system? (my own experience says no - converting an FRL-based program to an RLL-based one is not easy!). Are there problem domains in which one view is distinctly superior to the other? If so, what are they, and why is that view superior? Any answers or insights will be greatly appreciated... stan shebs ------------------------------ Date: Wed 27 Jun 84 11:35:48-PDT From: Michael Walker Subject: AI & statistics Ken, Thank you for mentioning our work on RADIX in the recent AILIST response about AI and regression analysis. It prompted me to put together a partial list of articles in AI and statistics, which I have been meaning to do. I've left out a number of articles by these authors in more obscure journals and proceedings. There is also work going on at Brunel University, and at BBN, but I haven't seen any publications from them yet. If people have additions to make, I would be happy to collect them and send them to the list. If readers would like reprints, the following addresses may be useful. Daryl Pregibon and Bill Gale can be reached at: Bell Laboratories 600 Mountain Avenue Murray Hill, New Jersey 07974 For D. Rodbard, write: D. Rodbard, M.D. National Institute of Child Health and HUman Development National Institutes of Health Bethesda, Maryland Our address here at the RADIX project is: Robert L. Blum and Michael G. Walker RADIX Project Department of Computer Science Margaret Jacks Hall Stanford University Stanford, California 94305 Mike Walker WALKER@SUMEX-AIM.ARPA [Blum 82a] Blum, R.L. Discovery and Representation of Causal Relationships from a Large Time-oriented Clincal Database: The RX Project. Springer-Verlag, 1982. Vol. 19 in the Medical Informatics series edited by D.A.B. Lindberg and P.L. Reichertz. [Blum 82b] Blum, R. L. Discovery, Confirmation, and Incorporation of Causal Relationships from a Large Time-Oriented Database: The RX Project. Computers and Biomedical Research 15(2):164-187, 1982. [Blum 82c] Blum, R. L. Induction of Causal Relationships from a Time-Oriented Clinical Database: An Overview of the RX Project. In Proceedings of the Symposium on Computer Applications in Medical Care. IEEE Computer Society, 1982. [Blum 84] Blum, R.L. Two-Stage Regression: Application to a Time-Oriented Clinical Database. 1984. in preparation. [Chambers 81] Chambers, J.M., Pregibon, D., and Zayas, E. Expert Software for Data Analysis: An Initial Experiment. In 43rd Session ISI. Buenos Aires, Argentina, 1981. [Gale 83] Gale, W.A., and Pregibon, D. Using Expert Systems for Developing Statistical Strategy. In Joint Statistical Meetings. Toronto, 1983. [Hajek 82] Hajek, P., and Ivanek, J. Artificial Intelligence and Data Analysis. In COMPSTAT 1982, pages 54-60. International Association for Statistical Computing, Physics-Verlag, Vienna, 1982. [Rodbard 83] Rodbard, D., Cole,B.R., and Munson,P.J. Development of a Friendly, Self-Teaching, Interactive Statistical Package for Analysis of Clinical Research Data: The BRIGHT STAT-PACK. In Seventh Annual Symposium on Computer Applications in Medical Care, pages 701-704. IEEE Computer Society, 1983. ------------------------------ Date: Wed, 27 Jun 84 09:36:41 PDT From: Michael Pazzani Subject: Spelling Correctors = Geography test correctors? Ignoring philosophical issues (after all, this is AILIST not a bad remake of "My Dinner With Andre") I don't feel that the spelling correctors or the geography test correctors are really that intelligent. The geography corrector seems to be very similar to the programs which grade SAT tests. Surely, one wouldn't want to call a SAT test correcting program AI even though it does a better and faster job than I would. I think its more important to discuss how to make these programs smarter. What would it take to have a spelling corrector find the intended word instead of all of the possibilities? A while ago, I worked on a program to do word sense selection. I wrote a spelling corrector for that program which treated a misspelled word as new word whose senses were the senses of all the possible corrections. It worked well when things like part of speech or selectional restrictions could disambiguate. How could one make this program smarter? Is it possible to try the "closer" possibilities first? Can you propagate the part of speech or semantic constraints into the search for possibilities? How would one store a large dictionary so it is efficient to find nouns, which are vehicles which look like "planh"? How can you detect a spelling error if the mistake is another word? (e.g. "I just typed rm *. Can you restore my flies from backup tape?) How do people do this anyway? ------------------------------ Date: 23 Jun 84 8:49:24-PDT (Sat) From: hplabs!hao!seismo!rochester!rocksvax!sunybcs!gloria!colonel @ Ucb-Vax.arpa Subject: Re: The Turing Test - machines vs people Article-I.D.: gloria.255 [This followup was actually written by a very clever computer program.] As you say, the Turing test is a _conversational_ test. Do you remember Turing's original "conversation"? "...Count me out on this. I never could write poetry." The whole conversation is fatuous! But then, it has no bonafide purpose. It was merely set up by a scientist to prove something. Nothing would be easier, for that matter, than to program a computer to take part in what Berne calls "8-stroke rituals": Hi. Hi. How are you? Fine. How are you? Fine. Nice day, isn't it? Yes. Well, goodbye. Goodbye. But would you want to carry on such a conversation with a computer? One converses socially only with conversers that one knows to be people. Col. G. L. Sicherman ...seismo!rochester!rocksanne!rocksvax!sunybcs!gloria!colonel ------------------------------ End of AIList Digest ******************** 28-Jun-84 12:00:38-PDT,11124;000000000000 Mail-From: LAWS created at 28-Jun-84 11:57:46 Date: Thu 28 Jun 1984 11:52-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@SRI-AI US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V2 #82 To: AIList@SRI-AI AIList Digest Friday, 29 Jun 1984 Volume 2 : Issue 82 Today's Topics: Humor & Business - How Not to Buy a Hero-1, Seminars - HAM-ANS Natural Language System, Expert System for Medical Consultation, Expert Systems at Hewlett-Packard, Conferences - Logic Programming Symposium, Workshop on Language Generation ---------------------------------------------------------------------- Date: 27 Jun 1984 08:23:02-EDT From: kushnier@NADC Subject: How Not to Buy a Hero-1 From kushnier@NADC Tue Jun 5 08:44:53 1984 Date: 5 Jun 1984 08:38:27-EDT From: kushnier@NADC To: SURINA@AFSC-HQ, kushnier@NADC.ARPA Subject: Re: Small Computer Procurements HOW NOT TO BUY A HERO 1 By Ron Kushnier I am an Engineer, by trade. I have my Double-E. And When I saw the Hero-1, I knew it was for me. I gave the order to my boss And explained the application, He very quickly signed the thing With the wildest jubilation. The order went to Purchasing And was no sooner in the door, When panic struck the buyer screamed, "No one bought a Hero-1 before". The order came back down to me With a simple "DISAPPROVED" My dreams were smashed My hopes were dashed My plans had been removed. Now I could understand this If a Shoeshine Boy I'd be But I'm supposed to be An Engineer And work in R&D. ------------------------------ Date: 28 Jun 1984 08:15:48-EDT From: kushnier@NADC Subject: Another Toy Another Toy By Ron Kushnier Here's another toy That my husband wants to buy. I'll never know What good it is Or any reasons why. But HERO-1, my fate is doomed I see it very clear. The age of Personal Computer Pets Is very nearly here. ------------------------------ Date: Tue 26 Jun 84 11:29:16-PDT From: Emma Pease Subject: Seminar - HAM-ANS Natural Language System [Forwarded from the CSLI Newsletter by Laws@SRI-AI.] The following will take place on Friday, June 29 in the Ventura Conference room from 2:00 to 4:00 (followed by tea). THE DIALOG SYSTEM HAM-ANS: NATURAL LANGUAGE ACCESS TO DIVERSE APPLICATION SYSTEMS (H. Marburger, K. Morik, B. Nebel) -- St This talk will introduce the overall goals of the NL-System HAM-ANS (HAMburg Application-oriented Natural language System) which is currently being developed at the University of Hamburg. HAM-ANS encompasses three different application classes: natural language access to a vision system (traffic at a street crossing), to a relational database system (fishery data), and for guiding a competitive dialog with a client (hotel reservation situation). The system accepts typed input in colloquial German and produces typed German responses. The system's general architecture and knowledge sources will be introduced. USER MODELING, EVALUATION STANDARDS, AND DIALOGUE STRUCTURE -- THE HAM-ANS APPROACH (Katharina Morik) -- AI dialogue systems are now developing from question-answering systems toward advising systems. This includes: * structuring dialog * understanding and generating a wider range of speech acts than simply information request and answer * modeling the user's familiarity with the system, his/her state of knowledge about the domain, and his/her evaluation standards (goals) In this talk, first the field of user modeling is structured according to the different aspects of the user (familiarity, knowledge, evaluation). We may then, secondly, describe our ongoing work in this field and relate it to other approaches. User modeling in HAM-ANS is closely connected to dialog structure and dialog strategy. In advising the user, the system generates the verbalizes speech acts. The choice of the speech act is guided by the user profile and the dialog strategy of the system. ------------------------------ Date: Tue 26 Jun 84 11:55:27-PDT From: Ted Shortliffe Subject: Seminar - Expert System for Medical Consultation [Forwarded from the Stanford bboard by Laws@SRI-AI.] There will be a special seminar presented by Mario Fieschi from Marseilles on Tuesday, July 10, from 2:30-3:30pm in the TC-135 conference room at the medical school. Mario has done some interesting work on medical expert systems, and is spending a few months at MIT with Peter Szolovits (who was on his thesis committee). He will be visiting Stanford from July 9-11. ------------------------------- Speaker: Mario Fieschi, MD, PhD Affiliation: University of Marseilles, France Title: SPHINX: An Expert System for Medical Consultations Place: Room TC-135, Medical School Time: Tuesday, July 10, 2:30-3:30pm I will present an outline of the program SPHINX, designed for the definition of medical knowledge and construction of a rule-based system, currently being used in: . Therapeutic decisions : Application in diabetes . Diagnostic decisions : Application in jaundice . Tool for education : Application in jaundice ------------------------------ Date: Wed 27 Jun 84 09:46:18-PDT From: Juanita Mullen Subject: Seminar - Expert Systems at Hewlett-Packard [Forwarded from the Stanford SIGLUNCH distribution by Laws@SRI-AI.] SIGLUNCH DATE: Friday, June 29, 1984 LOCATION: Chemistry Gazebo, between Organic & Physical Chemistry TIME: 12:05 SPEAKER: Steven Rosenberg Hewlett-Packard Research Laboratories Palo Alto TOPIC: Expert Systems at Hewlett-Packard The Applications Technology Laboratory of HP Labs is engaged in developing "industrial strength" AI. As part of its contribution to this effort, the Expert Systems Department has engaged in various "experiments" to develop expert system prototypes. One such experiment involved the development of PICC, an expert system for diagnosing flaws in IC wafers during negative photolithography. This talk will discuss the development and status of PICC. Besides describing the technical aspects of PICC, I will explore some of the issues involved in conducting expert systems experiments: why was photolithography chosen as a good area to apply expert systems technology; what were the pitfalls in moving PICC from a laboratory environment into a real fab line; even if it works, is it useful? ------------------------------ Date: 26 Jun 84 14:28:00-PDT (Tue) From: hplabs!hp-pcd!uoregon!conery @ Ucb-Vax.arpa Subject: Logic Programming Symposium Article-I.D.: uoregon.30100001 >From John Conery (conery@uoregon) -- Announcing -- 1985 International Symposium on Logic Programming Tentatively scheduled for Boston, Massachusetts, June 1985 Sponsored by IEEE Technical Committee on Computer Languages The symposium will cover implementations and applications of logic programming systems, including (but not limited to) parallel processing, expert systems, natural language processing, systems programming, implementation techniques, and performance issues. Authors should send 8 copies of their papers (8-20 pages, double spaced) to John Conery Department of Computer and Information Science University of Oregon Eugene, OR 97403 Submission deadline is November 1, 1984. A formal call for papers will be issued shortly. For more information, contact: Conference Chairman: Doug DeGroot IBM T.J. Watson Research Center PO Box 281, Yorktown Hts. NY 10598 Technical Co-Chairmen: Jacques Cohen Computer Science Dept - Ford Hall Brandeis University 415 South St Waltham MA 02254 CSNET: jc@brandeis ARPANET: jc.brandeis@csnet-relay John Conery Department of Computer and Information Sci University of Oregon Eugene, OR 97403 CSNET: conery@uoregon ARPANET: conery.uoregon@csnet-relay ------------------------------ Date: Thu 28 Jun 84 09:15:49-PDT From: Dikran Karagueuzian Subject: Workshop on Language Generation [Forwarded from the CSLI bboard by Laws@SRI-AI.] INTERNATIONAL WORKSHOP ON LANGUAGE GENERATION Organizers - Doug Appelt and Ivan Sag Staff - Emma Pease Dates - July 8 - 11 Size - 30 invited + 30 local Location - Stanford University Sponsors - National Science Foundation, American Association for Artificial Intelligence, CSLI, Fujitsu Laboratories, Ltd. The Second International Workshop on Language Generation will be held at Stanford University from July 8-10, immediately following the COLING conference. The workshop, organized by Doug Appelt and Ivan Sag is designed to allow researchers working in the field of language generation to share recent research results and discuss matters of importance to the field. Topics of discussion for this workshop include the design of grammatical formalisms for language generation, the role of planning and speech act theory in language generation, the production of extended discourse, the foundations for a theory of language generation, modeling the hearer's knowledge and intentions, and producing coherent explanations of reasoning and decision-making. Linguists as well as artificial intelligence researchers will participate in the workshop. The workshop is being sponsored by a grant from the National Science Foundation, the American Association for Artificial Intelligence, and a gift from Fujitsu Laboratories, Ltd. ADDITIONAL INFORMATION: Conference starts at noon, July 8 in the Elliott Program Center. This is a workshop and so interested people should check with Doug Appelt before going. [...] ------------------------------ End of AIList Digest ******************** authors point out that certain problems are intractable if dealt with symbolically, whereas they are easily solved if one uses real numbers and ordinary math. I suspect that the human brain uses a combination of analog and digital/symbolic processing, and that some cases of intuition might arise from the results of an analog computation into which introspection is not possible. As for Ken Laws's comment about switching to a new optimal strategy at each step (rather than Berliner's smoothing of transitions), one of the things he is trying to get around is the "horizon effect", where the existance of a sharp cut-off in the program's evaluation makes it think that postponing a problem solves it (since you no longer see the problem if it is pushed back over your horizon). In other words, perhaps the optimal strategy at each point *is* a non-linear combination of several discrete strategies. Also, I think it is a mistake to say that "pattern-matching" and "reasoning" are different things. After all, one must pattern-match in order to find appropriate objects to combine with an inference rule (obvious in OPS5, but also true in PROLOG). The question at hand is perhaps more whether one is allowed to use logically unsound inferences (a.k.a. heuristics). ------------------------------ Date: Mon 25 Jun 84 08:10:45-PDT From: PEREIRA@SRI-AI.ARPA Subject: ``Mind and brain'' mumbo-jumbo > From: Michael Dyer > The task of AI researchers > is to show how such vague notions CAN be understood computationally, > not to go around arguing against this simply because such notions > as "intuition" are so vague as to be computationally useless at > such at a bs level of discussion. It's like my postulating the > notion of "radio" and then looking at each transistor, crystal, wire or > what-have-you inside the radio, and then saying "THAT part can't be a > radio; that OTHER part there can't be one either. Just so! > From: hplabs!hao!seismo!rochester!ritcv!ccivax!band @ Ucb-Vax.arpa > Is it possible that "intuition" is the word we > use to explain what cannot be explained more > formally or logically? Why do these discussions always degenerate into suggestions of absolute limits to reason, perception or what not? That the task is *very* difficult we know, but we should not claim (without proof) that something *cannot* be done just because we cannot see how it could be done (within our lifetime...). Reminds me of those old ``if God had intended man to fly...'' arguments... Let's replace those ``what *cannot* be explained'' by ``what we can't yet explain''! -- Fernando Pereira pereira@sri-ai ------------------------------ Date: 25 Jun 84 16:27:57 EDT From: BIESEL@RUTGERS.ARPA Subject: Philosophy and other amusements. Judging from the responses on this net, the audience is evenly split between those who consider philosophy a waste of time in the context of AI, and those who love to dig up and discuss the same old chestnuts and conundrums that have amused amateur philosophers for many years now. First, any AI program worthy of that apellation is in fact an implementation of of philosophical theory, whether the implementer is aware of that fact or not. It is unfortunate that most implementers do *NOT* seem to be aware of this. Take something as apparently clear and unphilosophical as a vision program trying to make sense out of a blocks-world. Well, all that code deciding whether this or that junction of line segments could correspond to a corner is ultimately based on the (usually subconscious) presumption that there is a "real" world, that it exhibits certain regularities whether perceived by man or machine, that these regularities correspond to arrangements of "matter" and "energy", and that some aspects of these regularities can and should serve to constrain the behavior of some machine. There are even more buried assumptions about the time invariance of physical phenomena, the principle of causation, and the essential equivalence of "intelligent" behavior realized by different kinds of hardware/mushware (i.e. cells vs. transistors). ALL of these assumptions represent philosophical positions, which at other times, and in other places would have been severely questioned. It is only our common western heritage of rationalism and materialism that cloaks these facts, and makes it appear that the matter is settled. The unfortunate end-effect of this is that some of our more able practitioners (hackers) are unable to critically examine the foundations on which they build their systems, leading to ever more complex hacks, with patches applied where the underlying fabric of thought becomes threadbare. Second, for those who are fond of unscrewing the inscrutable, it should be pointed out that philosophy has never answered any fundamental questions (i.e. identity, duality, one vs. many, existence, essence etc. etc.). That is not its purpose; instead it should be an attempt to critically examine the foundations of our opinions and beliefs about the world, and its meaning. Take a real hard look at why you believe that "...Intuition is nothing more than..." thus-and-such, and if you come up with:'it is intuitively obvious', or 'everybody knows that', you've uncovered a mental blind spot. You may in the end confirm your original views, but at least you will know why you believe what you do, and you will have become aware of alternative views. Consider a solipsist AI program: philosophically unassailable, logically self-consistent, but functionally useless and indistinguishable from an autistic program. I'm afraid that some of the AI program approaches are just as dead-end, because they reflect only too well the simplistic views of their authors. Pete BIESEL@RUTGERS.ARPA (quick, more gasoline, I think the flames are dying down...) ------------------------------ End of AIList Digest ******************** 28-Jun-84 11:47:12-PDT,16273;000000000000 Mail-From: LAWS created at 28-Jun-84 11:43:22 Date: Thu 28 Jun 1984 11:38-PDT From: AIList Moderator Kenneth Laws Reply-to: AIList@SRI-AI US-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList Digest V2 #81 To: AIList@SRI-AI AIList Digest Thursday, 28 Jun 1984 Volume 2 : Issue 81 Today's Topics: AAAI - Instructions, Standards - Maintaining High Quality in AI Products, Business - Softwar, Mathematics - Best fitting curve, Knowledge Representation - Frames Que