nl-kr-request@cs.rpi.edu (NL-KR Moderator Chris Welty) (09/20/89)
NL-KR Digest (Tue Sep 19 18:38:21 1989) Volume 6 No. 37 Today's Topics: Better Explanations for Expert Systems (Unisys AI Seminar) Workshop Announcement & Call for papers Special Session on Logic and A.I. AI Seminar Re: Connectionist Approaches to NLU What KL-ONE Lookalikes Need to Cope with NL (Unisys Seminar) CALL FOR PAPERS Mathematical Programming and Expert Systems... Submissions: nl-kr@cs.rpi.edu Requests, policy: nl-kr-request@cs.rpi.edu Back issues are available from host archive.cs.rpi.edu [128.213.1.10] in the files nl-kr/Vxx/Nyy (ie nl-kr/V01/N01 for V1#1), mail requests will not be promptly satisfied. If you can't reach `cs.rpi.edu' you may want to use `turing.cs.rpi.edu' instead. BITNET subscribers: we now have a LISTSERVer for nl-kr. You may send submissions to NL-KR@RPIECS and any listserv-style administrative requests to LISTSERV@RPIECS. ----------------------------------------------------------------- To: nl-kr@cs.rpi.edu Date: Mon, 11 Sep 89 13:58:30 -0400 >From: finin@prc.unisys.com (Tim Finin) Subject: Better Explanations for Expert Systems (Unisys AI Seminar) AI SEMINAR UNISYS PAOLI RESEARCH CENTER Dr. Jennifer Jerrams Smith Artificial Intelligence Group Philips Research Laboratories Redhill, Surrey, RH1 5HA, UK (vrdxhq!uunet!prlb2!prlhp1!smithjj) The BEES project: Better Explanations for Expert Systems. The aim of the BEES project is to improve the communication and cooperation between users and knowledge based systems. It resulted from the realization that current expert systems (ES) provide very limited explanations to their users (the traditional "how" and "why" questions). In many cases their acceptability, marketability and usability would be greatly improved if such systems could interact in a more cooperative way to solve problems through the combined expertise of user and ES. Development of more cooperative ESs involves investigation of what domain knowledge is needed and how it should be represented so that the ES can answer some of the many different sorts of questions which users wish to ask. Additionally, the ES must hold information about the user, so that it can communicate effectively and efficiently, in terms which the user understands. We are currently working on the following areas: representing the variety of knowledge required, natural language processing, generating textual responses, integrating graphical and textual explanations, behavioural studies of advisors and users, developing a taxonomy of explanations. 11:00am September 25 BIC Conference Room Unisys Paoli Research Center Route 252 and Central Ave. Paoli PA 19311 -- non-Unisys visitors who are interested in attending should -- -- send email to finin@prc.unisys.com or call 215-648-7446 -- ------------------------------ To: nl-kr@cs.rpi.edu Date: Mon, 11 Sep 89 19:14:53 -0400 >From: kumard@cs.Buffalo.EDU (Deepak Kumar) Subject: Workshop Announcement & Call for papers CALL FOR PAPERS FIRST ANNUAL SNePS WORKSHOP Co-Sponsored by the SNePS Research Group and the SUNY at Buffalo Center for Cognitive Science November 13, 1989 280 Park Hall North Campus SUNY at Buffalo SNePS, in its various incarnations, is being actively used and developed at various AI research labs around the world. The aim of this workshop is to try to get together researchers who are (have been, or are considering) using SNePS as a research tool for AI modeling as well as those who are (have been, or are considering) evaluating SNePS as an AI research environment. The theme of this workshop is to present a survey of current and on-going research and developments at various research sites. Attendance at the workshop will be kept small (by invitation only) to allow for maximum possible interaction among participants. The workshop will be semi-formal and held on the campus of SUNY at Buffalo (where the SNePS headquarters are located). All papers presented will be edited and compiled into Proceedings which will be published as a Department of Computer Science, SUNY at Buffalo Technical Report. To be invited to the workshop, you have to do one of the following (preferably by e-mail): 1. Submit a one page abstract of a paper to be considered for presentation. 2. Submit a short write-up of your current research and how it does/might relate to the SNePS family of projects. We will also schedule a time slot for live demonstrations of systems. Those interested in giving demonstrations should submit a small description outlining the nature of the demonstration along with hardware/software/time requirements. DEADLINES Abstracts/Research Summaries due on Friday, October 6, 1989. Notification of acceptance/participation by October 13, 1989. Final papers due on Tuesday, October 31, 1989. SUBMISSIONS TO: Deepak Kumar 226 Bell Hall SUNY at Buffalo Buffalo NY, 14260 kumard@cs.buffalo.edu Send any queries to the organization committee: General: Deepak Kumar (kumard@cs.buffalo.edu) Sy Ali (syali@cs.buffalo.edu) Demos: Hans Chalupsky (hans@cs.buffalo.edu) Everyone will be expected to make their own living arrangements. There may be a possibility of making some arrangements with local SNeRG members. Thank you. Deepak Kumar ------------------------------ To: nl-kr@cs.rpi.edu >From: marek@gvax.cs.cornell.edu (Wiktor Marek) Newsgroups: comp.ai Subject: Special Session on Logic and A.I. Date: 12 Sep 89 17:15:08 GMT Reply-To: marek@cs.cornell.edu (Wiktor Marek) Special Session on Logic and Artificial Intelligence at: Symposium on Artificial Intelligence and Mathematics Fort Lauderdale, FL January 3-5 1990 During the Symposium (see: news.announce.conferences) a spe- cial Session on Logic and Artificial Intelligence will take place. A number of logicians currently involved in investi- gations of Logical Foundations of Artificial Intelligence will take part in the Session. These currently include: H. Blair (Syracuse U.), A. Brown (Xerox), M. Gelfond (U. of Texas), W. Marek (Cornell and U. of Kentucky), A. Nerode (Cornell), J. Schlipf (U. of Cincinnati), M. Truszczynski (U. of Kentucky), D. Wijesekera (Cornell). Scientists active in the research of Logical Foundations of Artificial Intelligence and interested in participating in the Session, are requested to get in touch with: Professor Anil Nerode Department of Mathematics Cornell University Ithaca, NY 14850 Electronic inquiries can be sent to: nerode@mssun7.msi.cornell.edu Wiktor Marek normally in the University of Kentucky, but now for 1989/90 in the MSI/ Cornell University. If you want to reach me, write to: marek@gvax.cs.cornell.edu ------------------------------ To: nl-kr@cs.rpi.edu Subject: AI Seminar >From: "Damaris M. Ayuso" <dayuso@BBN.COM> Reply-To: dayuso@BBN.COM Date: Thu, 14 Sep 89 14:45:00 EDT BBN STC Science Development Program AI Seminar Series Lecture AUTOMATIC PROCESSING OF UNRESTRICTED NATURAL LANGUAGE TEXTS GERARD SALTON Cornell University BBN STC, 2nd floor large conference room 10 Moulton St, Cambridge MA, 02138 Thursday September 21st, 10:30 AM The conventional artificial intelligence approaches are not viable when large text samples in arbitrary subject areas must be processed. Text analysis methods based largely on statistical methods have, however, been used successfully in information retrieval and related areas for many years. The main text analysis methods usable with general purpose texts are briefly reviewed and applications are described in such areas as document retrieval, book indexing, linking of related text excerpts, subject area organization, and so on. Evaluation output is shown wherever possible to demonstrate the effectiveness of the proposed methodologies. ------------------------------ To: nl-kr@cs.rpi.edu >From: munnari!ait.trl.OZ.AU!jacob@uunet.UU.NET (Jacob Cybulski) Newsgroups: comp.ai.digest,comp.ai.nlang-know-rep,aus.ai Subject: Re: Connectionist Approaches to NLU Date: 14 Sep 89 00:20:31 GMT Last week I sent a request for references on connectionist approaches to natural language processing. Special thanks to: Ron Chrisley of Xerox Trent Lange of AI Labs, UCLA Diana Roberts of HP Labs Here are some of the references they suggested, plus a few I found myself. Note that I also included approximate location of the authors. I'll be grateful if you could still contribute to this list and send me some additional information on yours or your colleagues' work in this area. Again I will summarize the responses to the net. Jacob o | o \_/ References: Bookman, L.A. (Brandeis Uni - 1987): "A Microfeature Based Scheme for Modelling Semantics," IJCAI 87, 611-614. Cottrell, G.W. and Small, S.L. (Uni of Rochester - 1983): "A Connectionist Scheme for Modelling Word Sense Disambiguation," Cog. and Brain Theory, 6(1), 89-120. Dolan, C. P. and Dyer, M. G. (UCLA - 1988): "Parallel Retrieval of Conceptual Knowledge," In D. Touretzky (ed), Proc. of the 1988 Connectionist Summer School, Morgan Kaufmann. Dyer, M. G. (UCLA - 1989): "Symbolic NeuroEngineering for Natural Language Processing," To appear in J. Barnden and J. Pollack (eds), Advances in Connectionist and Neural Computation Theory, Ablex Publications. Elman, J.L. (UCSD - 1989): "Representation and Structure in Connectionist Models," Cognitive Science Society Conference (not in proceedings) Elman, J.L. (UCSD - 1988): "Discovering Syntactic Structure Using Simple Recurrent Networks," Technical Report. Howells, T. (Uni of Massachusetts - 1988): "VITAL: A Connectionist Parser," Proc. Conf. Cog. Sci. 88, 18-25. Jones, M.A. (AT&T Bell Labs - 1987): "Feedback as a Coindexing Mechanism in Connectionist Architectures," IJCAI 87, 602-610. Lange, T. and Dyer, M. G. (UCLA - 1989): "Dynamic, Non-Local Role-Bindings and Inferencing in a Localist Network for Natural Language Understanding," In D. Touretzky (ed), Advances in Neural Information Processing Systems I, Morgan Kaufmann, San Mateo, CA. Lange, T. and Dyer, M. G. (UCLA - 1989): "Frame Selection in a Connectionist Model of High-Level Inferencing," Proc. of the Eleventh Annual Conf. of the Cognitive Science Society, Ann Arbor, MI, Aug 1989. MacWhinney, B. (Carnegie Mellon University - 1989): "A Crosslinguistic Connectionist Model for Morphological Learning," Cognitive Science Society Conference (not in proceedings) Marchman, V. & Plunkett, K. (UCSD & Uni of Aarhus, Denmark - 1989): "U-Shaped Learning Curves in a Connectionist Model of Past Tense Learning," Cognitive Science Society Conference (not in proceedings) McClelland, J.L. and Rumelhart, D.E. (Carnegie Mellon - 1981): "An Interactive Activation Model of Context Effects in Letter Perception," Psychological Reviews 88, 375-407. McClelland, J.L. (Carnegie Mellon - 1989): "Models of Language: Rules or Connections?" Cognitive Science Society Conference (not in proceedings) Miikkulannen, R. and Dyer, M. G. (UCLA - 1988): "Encoding Input/Output Representations in Connectionist Cognitive Systems," In D. Touretzky (ed), Proceedings of the 1988 Connectionist Summer School, Morgan Kaufmann. Miikkulainen, R. and Dyer, M. G. (UCLA - 1989): "A Modular Neural Network Architecture for Sequential Paraphrasing of Script-Based Stories," Proceedings of the First Annual International Joint Conference on Neural Networks, Washington DC, June 1989. Pollack, J.B. and Waltz, D.L. (Uni of Illinois at U-C - 1984): "Parallel Interpretation of Natural Language," Proc. Int. Conf. 5th Gen. Comp. Sys., 686-691. Small, S.L. (Uni of Rochester - 1987): "A Distributed Word-Based Approach to Parsing," in L. Bolc (ed) Natural Language Parsing Systems, Springer-Verlag, 161-202. St. John, M. (Carnegie Mellon University - 1989): "Sentence Comprehension by Parallel Constraint Satisfaction," Cognitive Science Society Conference (not in proceedings) ------------------------------ To: nl-kr@cs.rpi.edu Date: Tue, 19 Sep 89 11:33:15 -0400 >From: finin@PRC.Unisys.COM Subject: What KL-ONE Lookalikes Need to Cope with NL (Unisys Seminar) AI SEMINAR UNISYS PAOLI RESEARCH CENTER What KL-ONE Lookalikes Need to Cope with Natural Language Juergen Allgayer University of Saarbrucken FRG One of the major drawbacks of current NL processing systems is an adequate representation of and reasoning about plurals. This is true because current knowledge representation languages do not provide well suited representational means neither to describe sets, subsets, and elements nor to deal with the respective relations or use them in specially tailored inference systems. On the other side, there exists (at least) one linguistic theory about plurality in natural language, the General Quantifier Theory (GQT). What we want to present in this paper is how we adopted this theory into the already existing framework of the XTRA system. Our goal therefor is to develop a well-grounded knowledge representation formalism able to represent sets as well as to deal with them and combine this representation formalism with a well-defined linguistic theory. The knowledge representation language SB-ONE+ integrates sets into the KL-ONE like KR language SB-ONE. It realizes this by means of regarding sets as epistemological primitives, thus allowing for both an implementation of set-relevant properties (like reasoning about subset-of and element-of relationships) in the system as well as a description of sets as elements inside the TBox if relevant for the domain under consideration. Taking SB-ONE+ as representational basis, we show how some inte- resting results from GQT are implemented in the XTRA system. 11:00am September 20 BIC Conference Room Unisys Paoli Research Center Route 252 and Central Ave. Paoli PA 19311 -- non-Unisys visitors who are interested in attending should -- -- send email to finin@prc.unisys.com or call 215-648-7446 -- ------------------------------ To: nl-kr@cs.rpi.edu >From: E3T@PSUVM.PSU.EDU Newsgroups: comp.ai.nlang-know-rep Subject: CALL FOR PAPERS Mathematical Programming and Expert Systems... Date: Tuesday, 19 Sep 1989 12:38:37 EDT +-------------------------------------------------------------------+ | CALL FOR PAPERS | | | | RECENT APPLICATIONS OF MATHEMATICAL PROGRAMMING TO | | EXPERT SYSTEMS DEVELOPMENT | | | | A Section of the 8th INTERNATIONAL CONGRESS OF CYBERNETICS | | AND SYSTEMS | | | | Hunter College of the City University of New York | | New York, New York | | June 11-15, 1990 | +-------------------------------------------------------------------+ This session will host work that demonstrates the increasing potential of combining mathematical programming approaches to expert systems problems. Both original research and survey papers will be considered. Some relevant topics include, but are not limited to: o Machine Learning and Mathematical Programming o Logic Problems and Mathematical Programming o Expert Systems and Operations Research o AI and Operations Research This triennial conference is supported by many international groups concerned with management, the sciences, computers, and information systems. The Congress will provide a forum, symposia and sections, for the presentation and discussion of current research. All meetings will be held in midtown Manhattan. Section Chairs: Allen L. Soyster, Professor and Head, Dept. of Industrial Engineering, Penn State University Evangelos Triantaphyllou, Ph.D. Candidate, Dept. of Industrial Engineering, Penn State University Program Chair: Constantin Negoita, Professor, Hunter College, CUNY, New York, NY. Papers should be approximately 2,000-4,000 words in length. Please send 4 hard copies (not e-mail) to: +-------------------------------------------------------------------+ | Evangelos Triantaphyllou | | E-Mail: E3T@PSUVM.BITNET | | Department of Industrial Engineering | | Penn State University | | 207 Hammond Building | | University Park, PA 16802, USA | +-------------------------------------------------------------------+ Deadlines: January 1, 1990: Submission of paper. April 1, 1990: Notification of disposition All items will be acknowledged ------------------------------ End of NL-KR Digest *******************