Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!haven.umd.edu!news.umbc.edu!eff!news.duke.edu!agate!howland.reston.ans.net!swrinde!cs.utexas.edu!uunet!in2.uu.net!mr.net!news.mr.net!timbuk.cray.com!walter.cray.com!awp
From: awp@cray.com (Adrian Powell)
Subject: The Best natural language sys
Message-ID: <1996Jan4.042949.18744@walter.cray.com>
Lines: 20
Nntp-Posting-Host: hydra-eth.cray.com
X-Newsreader: TIN [version 1.2 021193BETA PL3-CRIc]
Date: 4 Jan 96 04:29:48 CST

Hi,
     Back in the mid eighties, I looked at ai for sometime, mainly in
the area of natural language processing and machine understanding. Although
soem interesting concepts and ideas were around, we lacked large amounts of
computer power, memory etc. Seeing the obvious shortfall, I havn't looked 
back at this area until now. Half expecting great leaps to have been made,
but supprisingly, no. Not much appears to have happened. 
    Some months ago, I did see a program on Television about a team in the US
who where working on a seemingly quite powerful natural language system. They
appeared to feed it information in the daytime, and left it over night to 
try to cross reference and assimulate all the information that it had learn't
that day, into its knowledge base. This gave it what seemed to be a good
learning capability.  Does anyone know of information on such a system ?. 
At the time I really didn't take much notice where the team were working at ?.
      What the capabilities of the best state of the art program in this
field currently ?.

      Many thanks,   Adrian Powell


