IRList Digest Sunday, 1 November 1987 Volume 3 : Issue 37 Today's Topics: Reply - Personal filing systems (offer to help) Query - Software to assist scholars of ancient texts - Relational DB toolkits with source code - Network re social systems science - How types of knowledge are classified (esp. in organizations) - Neural networks and IR - Hypertext discussion group Discussion - Literature on browsing Announcement - Conference on the Uses of Large Text Databases Call for Papers - CAIS (linking mind and machine) News addresses are Internet or CSNET: fox@vtcs1.cs.vt.edu BITNET: fox@vtcs1.bitnet ---------------------------------------------------------------------- Date: Mon 5 Oct 87 10:05:39-PDT From: Ken Laws Subject: Research on Filing Systems [See V3 #36 - Ed] Cc: IIN8PXB%UCLAMVS.BITNET@WISCVM.WISC.EDU, STEIN@WHARTON.UPENN.EDU A very interesting IRList issue. While I have no expertise in the theory of personal filing systems, I have always been fascinated by taxonomies and other means of organizing knowledge. I would be happy to help out if someone wants to study the evolution (or current structure) of the machine vision field. (This includes digital image processing, analysis, and understanding, with interfaces to perceptual psychology, neurophysiology, computational geometry, remote sensing, geology, geodessy, computer graphics, computer science, mathematics, and many other areas. I can't claim to understand it all -- indeed, computer vision is itself not well understood -- but I am familiar with much of the literature.) -- Ken Laws ------------------------------ Date: Sun, 4 Oct 1987 18:52 CST From: Robin C. Cover Subject: texts of ancient languages Professor Fox: I am a professor of semitics and am interested in full-text retrieval software which can be used to concord/index/search machine-readable texts of ancient languages. I am also interested in various aspects of mechano- linguistics: statistical thematics, content analysis, disambiguation, etc. I would appreciate being added to the mail list for (BITNET) IRList -- in hopes of making contact with other scholars interested in study of ancient texts. Many thanks! Professor Robin C. Cover; 3909 Swiss Avenue; Dallas, TX 75204 214/824-3094. Acknowledge-To: Robin C. Cover PS. I would like to be added to your IRList. The topics given in the ARPANET SIGLIST sound like it's the SIG most centrally focused on my area of interest. I am helping supervise a hypertext project, and we need advice at critical junctures from people in AI, linguistics and similar sub-disciplines. ------------------------------ Date: Thu, 15 Oct 87 10:58:10 PDT From: Mark Richer Subject: Relational DB toolkits I'm interested in finding RDB toolkits with full source provided for programmers. Multi-user capabilities would be very nice. I should say I'd like source in C. I want to use this on a Macintosh so it has to be possible to port. I know of one toolkit for the Mac called inside out which comes as a linkable object library, but they won't sell source and I don't know anything about the product (do you?). I'm not even sure they have anything execpt the MPW Pascal version shipping, but they claim they will support a host of C and Pascal compilers on the Mac. Thanks in advance for any comments. Mark ------------------------------ Date: Fri, 2 Oct 87 19:40 EST From: STEIN@wharton.upenn.edu Subject: network for those interested in social systems sciences . . . Finally, I wonder if I can create a network for that subset of the irlist (or other lists) who are interested in research in social systems sciences coming from the same tradition as my graduate group (exlained in previous memo)--please advise. . . . Sincerely Yours, Eric W. Stein Social Systems Sciences The Wharton School U. Penn Stein@Wharton.upenn.edu [Note: I certainly have no objects to others setting up discussion groups, but am happy to include any subscribers and news in this area if it comes my way. Eric is referring to his msg in V3 #36 in case you have forgotten - sorry for not getting this follow up msg out sooner. - Ed] ------------------------------ Date: Tue, 20 Oct 87 21:26 EST From: STEIN@wharton.upenn.edu Subject: knowledge classification . . . I'm interested in how types of knowledge are classified; e.g. declarative, procedural, implicit, tacit, etc. Does anyone know the seminal works in this area? Most current research, especially in AI and in the expert database systems literature? I'm interested in the types of knowledge structures that exist in organizations. yours, Eric Stein U. Penn ------------------------------ Date: Wed, 28 Oct 87 12:12:46 PST From: richardp@violet.Berkeley.EDU Subject: Neural networks and IR Richard Petersen School of Library and Information Studies 2150 Kittredge St. Berkeley, CA. 94720 (415) 642-1464 1210 Paru St. #E Alameda, CA. 94501 (415) 523-9571 richardp@violet.bitnet Dear Professor Fox, I am currently a doctoral student at the School of Library and Information Studies at U.C. Berkeley. My field of research is the application of Neural Networks (also know as Parallel Distributed Processing or Connectionism) to the problem of Information retrieval. I would be interested in contacting and corresponding with anyone who is currently working on similar kinds of applications. I have also heard that you are currently compiling a collection of test data bases used for information retrieval studies. I would be extremely interested in the possibility of acquiring such test retrieval systems for my research. Of particular interest are those retrieval data bases used by Salton and Spark-Jones; Cranfield 400, Smart, and Inspec. [Note: I plan on having a number on a CDROM. I encourage IR researchers, especially those in Europe who have collections that I do not have, to contact me so NPL, Cranfield, etc. can be included. - Ed] Such a retrieval data bases may serve as a comparative benchmark of a neural networks effectiveness in terms of other information retrieval systems. Neural Networks are characterized by binary vector and matrix representations similar to that used in document-term matrices. The term based binary representation of documents may be formulated as a distributed representation (parallel distributed pro- cessing) used in neural. networks. Such a representation could be used in classification and feedback processes employed in neural networks. My particular area of study is in the application of relevance feedback mechanisms to ther- modynamic PDP models such as the Boltzman machine and Har- mony theory. I would appreciate any help you could give me, espe- cially in regard to the acquisition test retrieval systems. I would also greatly appreciate your evaluation on the use of such retrieval data bases in the application of Neural Networks to the Information Retrieval problem. Sincerely, Richard Petersen richardp@violet.bitnet [Note: I will make an announcement regarding the CDROM with IR collections when dates and prices are known. I suggest you contact Bruce Croft at croft%cs.umass.edu@relay.cs.net regarding your work since there may be a conference in part relating to this. - Ed] ------------------------------ Date: Mon Aug 31 16:35:16 1987 From: omepd!littlei!foobar!sdp!sdp@seismo.CSS.GOV Subject: Hypertext Hello, . . . I'm looking for a program called HYPE. It's a small hypertext editor written in Turbo Pascal. Anyone know of any other interesting PD hypertext code? Is there a hypertext mailing list or discussion group anywhere on the net? Is there demand for one? Thanks, Scott Peterson OMO Software Intel, Hillsboro OR sdp.hf.intel.com!sdp omepd.intel.com!littlei!foobar!sdp!sdp [Note: I found this msg in AIList Digest and wonder if there were any replies about the discussion group. Hypertext certainly seems to me to be a topic that should be covered by IRList -- I welcome any news and discussion. See the following message, for example. I certainly hope that we get a number of reports from people going to Hypertext '87, which will occur very shortly. - Ed] ------------------------------ Date: Tue, 27 Oct 87 13:11:23 CST From: JEFF HUESTIS Subject: QUICK BIB ON BROWSING I've tapped into an interesting part of the database literature lately, starting from citation number 29 in Salton's selected abstracts in SIGIR Forum 21(3-4). The concept of browsing is critical for anyone trying to model user behavior, and the small body of database browsing literature seems to offer some useful terminology and ways of looking at things that are applicable to IR. There doesn't seem to be a whole lot out there, so it may be useful to someone if I list a couple of representative citations, aside from the above. If this is old hat to anyone, I'd appreciate contributions. Larson, James A. "A visual approach to browsing in a database environment," Computer, 19(6):62-71; June 1986. ^^^^^^^^ Motro, Amihai. "Browsing in a loosely structured database," SIGMOD Record, 14(2):197-207; 1984 (Proceedings of the SIGMOD ^^^^^^ ^^^^^^ Annual Meeting, Boston, June 18-21, 1984, edited by B. Yormark). A common thread through all this stuff seems to be the acknowledgement that not every user knows what he wants, but he knows what he likes when he sees it. Adaptive search strategies, relevance feedback, query refinement, etc. The ideal IR system should work from the initial query: "Uh..." [Note: In some ways, the concept of hypertext is simply that of intelligent, personal or aided, browsing. I feel that there should not be a split into camps of people working on hypertext and others working on IR -- we will all benefit if hypertext capabilities are included in those IR systems where such is possible. After all, V. Bush's memex inspired both kinds of effort. - Ed] ------------------------------ Date: Wed, 7 Oct 87 10:00:58 EDT From: Maureen Searle Subject: Conference reminder Would it be possible to include this announcement in the IRList DIGEST? Thank you. Maureen Searle Centre for the New OED University of Waterloo msearle%watsol@waterloo.csnet ______________________________________________________________________ University of Waterloo, Centre for the New OED Waterloo, Ontario Conference on the Uses of Large Text Databases November 9th and 19th, 1987 Deadline for registration is October 15. Because of a rotating Canadian postal strike, registrations can be made by phone (519) 885-1211 ext 6183 or by electronic mail to mjfehlner%watsol@waterloo.csnet. Conference fees are:- Registration $90.00 (CDN) $68.00 (U.S.) Proceedings $20.00 (CDN) $15.00 (U.S.) Mon. evening Dinner $30.00 (CDN) $23.00 (U.S.) Recommended hotel is the Walper Terrace Hotel, 1 King Street West, Kitchener, Ontario, N2G 2Z9. Rates are $54.00 (CDN)/night (single) and $60.00 (CDN)/night (double). Reservations should be made directly with the hotel by calling (519) 745-4321, specifying that you are attending the conference. Transportation from Toronto Airport to Kitchener is available from Airways Transit at a special conference rate of $17.00 (one way). Reservations can be made by calling (519) 578-0110, again specifying that you are attending the conference. If you have further questions please contact the Centre at the above number or via electronic mail. ------------------------------ Date: 21 Oct 87 20:04 -0300 From: Michael Shepherd Subject: CALL FOR PAPERS -- Connexions: linking mind and machine . . . CALL FOR PAPERS Connexions: linking mind and machine CANADIAN ASSOCIATION FOR INFORMATION SCIENCE The 16th annual conference of the Canadian Association for Information Science will convene at the University of Ottawa, in Ottawa, Canada, May 12 - 14, 1988. This conference will interest information professionals, data processing and computer specialists, engineers and others involved in documentation, communications and computer science. Previously unpublished papers are solicited. Authors are invited to submit papers for presentation in either English or French. Topics may include, but will not be limited to,the following: - the transfer of information through the information chain from source to supplier to consumer - new hardware technologies and software systems and how they connect to existing systems and services - new ways of organizing knowledge to expedite its transfer, such as artificial intelligence and expert systems - ergonomics and the studies of the interactions between information systems and users - the study of information science and its links to other disciplines as diverse as psychology, linguistics, communications, computer science and engineering Three copies of a 500 to 1000 word abstract with the author's full name, affiliation, address and telephone should be submitted to: Prof. Charles Meadow c/o CAIS 88 P.O. Box 38, Station B Ottawa, Ontario Canada K1P 6C3 Selected papers will be published in the Canadian Journal of Information Science. IMPORTANT DATES: 15 Nov - Abstract Due 30 Dec - Papers Due For further information regarding the conference, please contact: David Holmes at the address above, or by telephone: (613)-564-4074 ------------------------------ END OF IRList Digest ********************