Newsgroups: comp.ai.nat-lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!swrinde!hookup!olivea!uunet!newsflash.concordia.ca!news.mcgill.ca!clouso.crim.ca!athena.ulaval.ca!ppp8.ulaval.ca!user
From: gauvins@vm1.ulaval.ca (Stphane Gauvin)
Subject: text analysis
Message-ID: <gauvins-2212941150410001@ppp8.ulaval.ca>
Sender: news@athena.ulaval.ca
Nntp-Posting-Host: ppp8.ulaval.ca
Organization: Universit Laval
X-Newsreader: Value-Added NewsWatcher 2.0b22.0+
Date: Thu, 22 Dec 1994 16:53:00 GMT
Lines: 24

A colleague and I are doing research on group decision support systems and
idea generation. A typical issue is idea unicity. Experts must examine all
ideas, one by one, in order to determine what an individual has added to
the group's pool of ideas. 

Our current reserch focuses on psycho-social dimensions of idea
generation. We conduct repetitive experiments on identical tasks (e.g.
uses for a knife). We thought of automating the task of analyzing the
stream of ideas. Ideas are one-liners, and so far I believe that we have
identified a couple of hundreds of ideas (+/- 500 subjects).

I would appreciate your ideas/pointers on whether idea analysis can be
automated. Can a neural net be trained by an expert efficiently enough
such that all (or at least a large fraction) of the ideas could be
categorized?

Thanks

Stephane

-- 
Stephane Gauvin                                 gauvinS@vm1.uLaval.ca
Departement of Marketing                           (418) 656-2158 off
Universite Laval                                   (418) 656-2624 fax
