Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!elroy.jpl.nasa.gov!usc!samsung!uunet!sps!vjk
From: vjk@sps.com (Vince Kovarik)
Newsgroups: comp.ai
Subject: Re: NLP reference wanted
Summary: Syntactic differences are present....
Message-ID: <285@sps.com>
Date: 5 Dec 90 01:02:15 GMT
References: <25674@uflorida.cis.ufl.EDU> <8038@ucdavis.ucdavis.edu>
Organization: SPS - Indialantic, FL
Lines: 35

In article <8038@ucdavis.ucdavis.edu>, mock@iris.ucdavis.edu 
(Kenrick J. Mock) replies:

> 
> Semantics are necessary to really understand what is going on.
> Consider the following sentences:
> 
> "John gave Mary a book."
> "John gave Mary a kiss."
> "John gave Mary a beating."
> 
> All of these sentences are syntactically the same, but obviously mean
> different things.  
> 

Although a firm believer in the necessity of semantics, there is,
nonetheless, syntactic differences in the examples provided.

A book would be identified as a noun and, hence, some object which
is the target of the "giving" action.

Both kiss and beating are verbs but different forms (i.e. simple 
present and present participle).  

Now, although the purely syntactic parser might be able to recognize
these differences, it does nothing beyond the construction of simple
assertions that these different forms of the "give" event did occur.

To do something *USEFUL* these events need to be represented within 
a semantic framework that allows the system to remember and reason
about the events.  Certainly some default assumptions may be made
about the relationship between John and Mary based on whether the
object was a "kiss" or a "beating."  This type of inference is not
possible in syntactic-only approaches (and if someone says it is, then
they're building semantic representations but don't want to admit it).


Brought to you by Super Global Mega Corp .com