Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!usenet.eel.ufl.edu!bofh.dot!warwick!bham!sunserver1.aston.ac.uk!usenet
From: rupraibs <rupraibs@aston.ac.uk>
Subject: Re: Transition Network for English Grammar. Possible or not?
X-Nntp-Posting-Host: finland.aston.ac.uk
Content-Type: text/plain; charset=us-ascii
Message-ID: <DrrL8E.LI3@aston.ac.uk>
Sender: usenet@aston.ac.uk
Content-Transfer-Encoding: 7bit
Organization: Aston University
Mime-Version: 1.0
Date: Tue, 21 May 1996 16:45:50 GMT
X-Mailer: Mozilla 1.1N (X11; I; SunOS 5.4 sun4c)
X-Url: news:comp.ai.nat-lang
Lines: 27

I agree that there will be ambiguity for sentences such as:
Time flies like an arrow.
Fruit flies like a banana.
A recursive transition network may parse them in a manner that humans would
regard as incorrect in that "flies" could be interpreted as a verb in the
second sentence.
However, if a recursive transition network were being used that would have 
several ways to parse the sentence, it could report the above two sentences as
correct. Without actually being told the specific route taken through the
transition network to parse the sentences, a human may come to the conclusion
that the parser was doing its job correctly, no matter which route it had taken
(even if it had taken "flies" as a verb in the 2nd sentence).

This transition network would not return how it had parsed the sentence or even
all possible parses. It would just return "correct" or "incorrect".
Obviously this is a very loose interpretation of the meaning of "parse".

So, is it still possible to build a transition network to parse the whole of
the
English language. Note, I am making it easier and easier for the network to see
how far I can go. Can anyone think of a sentence that would break the network
using the above scheme or point out more flaws?

Thanks for all your comments so far.

Roop.

