Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!zombie.ncsc.mil!gmi!msuinfo!uchinews!iitmax!sanders
From: sanders@iitmax.iit.edu (Greg Sanders)
Subject: Re: Books on Intro. natural language proces
Message-ID: <1994Sep30.221823.12474@iitmax.iit.edu>
Organization: Illinois Institute of Technology / Academic Computing Center
References: <36ci4i$if6@narnia.ccs.neu.edu> <1994Sep29.063344.29297@cs.cornell.edu> <36f45q$rnb@narnia.ccs.neu.edu>
Date: Fri, 30 Sep 94 22:18:23 GMT
Lines: 31

In article <36f45q$rnb@narnia.ccs.neu.edu> hafner@ccs.neu.edu (Carole Hafner) writes:
>A quick clarification: 
>The research that was (to some extent) "abandoned" was not work on deep 
>parsing and linguistic theory, which goes on as usual, but rather the 
>integration of domain knowledge models with NLP systems, discourse analysis, 
>language as goal-directed behavior ("speech acts"), and other studies 
>of language from an AI point of view.

Oh! I think this is a a rather one-sided description of the state of NLP.
None of this stuff has been abandoned.  What has happened is more a matter
of seeing no quick easy triumphs from such research.  We now see that they
are larger and more difficult problems than was appreciated.  As a result,
there seems to be more willingness to draw on philosophy, linguistics,
conversation analysis, sub-language analysis, cognitive psychology, and 
so on.  Dr. Hafner is right that many researchers really are going with
alternative approaches, but *much* depends on the task you want to perform.

If you want to construct a dialogue, you have no choice but to study
speech-act sorts of analyses, and you *most*certainly*do*not* forget
about discourse analysis and generation.  Correspondingly, tutoring
systems involve a great deal of work on integrating domain knowledge
systems with NLP (both for understanding and generation).

Finally, I think that discourse analysis and generation is one of
the hottest topics in NLP and has been so for a few years.  Take a
look through recent years of Computational Linguistics for examples.
In my opinion, some real successes have resulted from that work.

-- Greg Sanders


