Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!pipex!uknet!festival!edcogsci!steve
From: steve@cogsci.ed.ac.uk (Steve Finch)
Subject: Wornet and IR (Was Re:Commercial NLP or Information Retrieval)
Message-ID: <D6IM3t.21J@cogsci.ed.ac.uk>
Organization: Centre for Cognitive Science, Edinburgh, UK
References: <3kaalq$n19@booz.bah.com><TED.95Mar23194654@hellespont.crl.nmsu.edu> 	<3l29ut$a2v@pipe4.nyc.pipeline.com> <TED.95Mar25220339@hellespont.crl.nmsu.edu>
Date: Tue, 4 Apr 1995 14:40:38 GMT
Lines: 32

ted@crl.nmsu.edu (Ted Dunning) writes:

>In article <3l29ut$a2v@pipe4.nyc.pipeline.com> mitioke@readware.com
>(Ken Ewell) writes:

>>   The author's state: "The results support the claim that systems
>>  incorporating natural language-processing techniques are more
>>  effective than systems based upon stochastic techniques alone."

>for MUC tasks.  not for information retrieval tasks.  and as a point
>of information, stochastic is a rough synonym for random.  a better
>term would be statistically based

>>   One of their conclusions was: "..the top-scoring MUC-3 systems
>>  incorporate a diverse range of natural language-processing
>>  techniques.  With so many different approaches demonstrating
>>  viability, long term prospects for information extraction based
>>  upon natural language processing are very promising."

>note the mention of information extraction.  i think you are seriously
>confused here.

On a related note, I have often heard claims to the effect that
"wordnet (synonym/hyponym thesaurus) doesn't help IR", and even recall
reading/hearing a paper to such effect.  However I don't recall the
reference and if anyone has such a reference I would very much
appreciate it.

Cheers,

Steve.

