Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!ucbvax!ulysses!poesio From: poesio@research.att.com (Massimo Poesio) Newsgroups: comp.ai Subject: Recent Natural Language Interfaces Message-ID: Date: 11 Jul 90 18:35:24 GMT Sender: netnews@ulysses.att.com Distribution: comp.ai Organization: AT&T Bell Laboratories Lines: 73 In-Reply-To: yvonne@gufalet.let.rug.nl's message of 10 Jul 90 08:09:14 GMT The posting that follows is a reply to a recent posting by Yvonne Vogelenzang asking about recent natural language interfaces. I tried to reply via email, but my message bounced. Since the content is of moderately general interest I'll try to followup instead - no reference to the Chinese Room is made, though. REST OF MESSAGE FOLLOWS ---------------------------- What do you mean by natural language interfaces? I'll assume that you are asking about systems which are able to understand natural language questions (and possibly reply in english too). You can find lots of information about recent natural language interfaces in the journal Computational Linguistics, as well as in the proceedings of the ACL; your Computer Science Dept. should have both. Especially useful would be to get the proceedings of the 1988 conference on Applied Natural Language Processing, where lots of systems where discussed. The most famous `recent' systems include: - TEAM, developed at SRI at the beginning of the '80s; its `final' description appeared in Artificial Intelligence Journal, 1987 - UC, still under development at Berkeley, by the group of Wilensky; a description appeared in Computational Linguistics in 1989 or 1988; - The system developed by Kathleen Dahlgren and Michael McCord of IBM: again, a description appeared in Computational Linguistics in 1988 or 1989, plus Kathy had a book in 1987 called `Naive Semantics' which is a great book; - XTRA, under development at University of Saarbruecken; a description appeared in the International Journal of Man Machine Studies; - LILOG, under development at the University of Stuttgart and IBM Germany. - LUCY, under development at MCC by the group of Elaine Rich; Len Schubert, has developed a story understanding system called ECOLOGIC which makes use of GPSG to build a Montague-like representation called Episodic Logic; a description of Episodic Logic will appear in Artificial Intelligence Journal shortly. The group of James Allen at Rochester is developing a discourse understanding system, but the work is still in its beginnings; again, I could tell you more since I am involved in this. Last but not least, one system I have worked on, WISBER, developed by the University of Hamburg, Siemens Munich et al. We had tons of technical reports (available by writing at the Department of Computer Science), plus papers at ECAI, COLING and GWAI. If you are in a hurry, start looking at the first systems I've mentioned, in that order. Hope this helps Massimo Poesio University of Rochester, Dept. of Computer Science Rochester, NY 14620 - USA poesio@cs.rochester.edu (currently at AT&T Bell Labs, 600 Mountain Ave. Murray Hill, NJ 07974 poesio@research.att.com)