emunix.emich.edu>
Editor for this issue: Annemarie Valdez <avaldezemunix.emich.edu>
CS.Wayne.EDU>Dear Colleagues, The submission deadline for the AAAI Fall 1996 Symposium on KNOWLEDGE REPRESENTATION SYSTEMS BASED ON NATURAL LANGUAGE is extended to May 7, 1996, Tuesday. We revise this deadline because of the discrepancy between two different dates, April 15 and May 1, that I mistakingly communicated to you. If you have already submitted your paper, you can either do nothing or send us your newest version by May 7. FYI, I am enclosing the cfp with some revised dates. Best wishes Lucja ======================================================================= KNOWLEDGE REPRESENTATION SYSTEMS BASED ON NATURAL LANGUAGE Call for participation in a AAAI Fall 1996 Symposium WHERE: Boston/Cambridge, MIT WHEN: November 9-11, 1996 shortly after KR-96 to be held November 4-7, 1996 http://www.cs.wayne.edu/nlkr - -------------------------------------------------------------------- The Symposium addresses the theoretically and practically important problem of knowledge representation (KR) systems that closely parallel the representational and inferential characteristics of natural language (NL). Advantages of such NL-based KR systems would be enormous. Among the arguments for the natural-language-as-KR-system approach are: 1. KR systems based on natural language would be easy for people to use, 2. Most human knowledge is encoded and communicated via natural language, in the form of textual documents and (transcribed) interactions (dialogs). A NL-based KR system would be capable of automatically creating and updating its knowledge base from natural language texts more easily. Additionally, the contents of this knowledge base and inferences supported by the KR system would parallel those of a natural language user. 3. Every day, a huge number of new textual documents becomes available on-line. This creates the need for more sophisticated information retrieval techniques basedn natural language processing (NLP) and KR techniques. 4. KR systems based on natural language would provide a uniform symbolic representation. The same representational and inference mechanism could be used when utilizing previous knowledge for processing new natural language inputs (natural language as both meta-level and object-level language), 5. It is hard to match expressiveness and precision of natural language, particularly in not (well) formalized domains, 6. Many philosophers, linguists and cognitive scientists believe that mental-level representation of knowledge (human mind) is close in form to natural language. While some AI researchers believe that it is feasible and necessary to design KR systems closely mimicking natural language, others are pessimistic about success or even possibility of designing such KR systems. This pessimism might account for the general lack of interest in the problems of NLP within the KR community; for example, only six of the twenty-two KR systems presented in the "Special Issue on Implemented Knowledge Representation and Reasoning Systems", Charles Rich, Editor SIGART Bulletin, Vol. 2 (3), ACM Press, 1991, are driven by NLP concerns. Among the arguments against the NL-as-KR-system approach are: 1. Natural language is (highly) ambiguous, 2. Natural language has (very) complex syntax, semantics, and pragmatics, 3. Natural language is non-systematic, non-algorithmic, 4. Natural language is (highly) context-dependent, 5. Natural language is (merely) an interface; Inferencing does not belong with natural language. The goal of this Symposium is to address in-depth such arguments for and against designing KR systems closely simulating natural language. We invite papers that substantiate the view that natural language can be viewed as a KR system with its own representational and inferential machinery, and, as such, is a productive source of ideas for KR formalisms and their practical implementations. We are interested in papers discussing representations and inference mechanisms paralleling a non-trivial or interesting subset of natural language and formal systems whose expressiveness, semantics, information packaging, reasoning, and computational tractability closely correspond to that of natural language. We are interested in automatic or semi-automatic methods of obtaining taxonomies facilitating various NLP tasks such as anaphora resolution, inferencing, and machine translation. We are also interested in papers that discuss those aspects of natural language that are not desirable in a KR system. We invite position papers with supported arguments against the idea of designing KR systems that mimic natural language. - -------------------------------------------------------------------- PAPER FORMAT: Strongly preferred 12 pt article latex style 15 pages maximum, including title, abstract, figures, but excluding references The first page must include: title author's name(s) affiliation complete mailing address e-mail address phone/fax number(s) abstract of 200 or so words keywords ELECTRONIC SUBMISSIONS are strongly preferred: DIRECT: anonymous ftp to ftp.cs.wayne.edu ~pub/nlkr directory As the last resort, five hard copies of the paper can be snail mailed to Lucja Iwanska Department of Computer Science Wayne State University Detroit, MI 48202, USA (313) 577-1667 (phone) (313) 577-2478 (secretary) (313) 577-6868 (fax) TIMETABLE: January 1, 1996 intent to submit due May 7, 1996 (earlier it was: April 15 and May 1, 1996) June 15, 1996 (earlier it was: May 17, 1996) reviews completed papers chosen notification/comments/requests for changes sent out August 23, 1996: camera-ready papers, signed "Permissions to Distribute" forms and A/V requests received by the chair November 9-11, 1996: Symposium takes place - ------------------------------------------------------------------------- PROGRAM COMMITTEE: Syed S. Ali, Southwest Missouri State University syaliMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue