Newsgroups: comp.theory.info-retrieval
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!news.sprintlink.net!hookup!ames!news.hawaii.edu!uhunix3.uhcc.Hawaii.Edu!pollarda
From: pollarda@uhunix3.uhcc.Hawaii.Edu (Art Pollard)
Subject: Re: Recall and Precision optimisation using GAs ?
X-Nntp-Posting-Host: uhunix3.uhcc.hawaii.edu
Message-ID: <D71MpE.Gww@news.hawaii.edu>
Sender: news@news.hawaii.edu
Organization: University of Hawaii
References: <D6zCBw.KKn@curia.ucc.ie> <3mmfkr$8rq@usenet.srv.cis.pitt.edu>
Date: Fri, 14 Apr 1995 21:08:02 GMT
Approved: Pollarda@Uhunix.uhcc.hawaii.edu
Lines: 31

In article <3mmfkr$8rq@usenet.srv.cis.pitt.edu> dsdst3@lis.pitt.edu (Dave Dubin) writes:
>Mike McElligott (mike@csws4.ucc.ie) wrote:
>
>: 	Does anyone know of any work done on using a genetic algorithm to try
>: and simultaneously optimise recall and precision in an IR model ?
>
>  Yes. 
>
>   "Query Modification Using Genetic Algorithms in Vector Space Models,"
>    Yang, Jing-Jye and Robert Korfhage
>    International Journal of Expert Systems
>    Pp. 165-191, Volume 7, Number 2, 1994. ISSN 0894-9077
>
>The article is based on J.J.'s dissertation work that he did here at Pitt. 
>I believe they also participated in TREC 1.


There was a simular article in the CACM some time ago.  The article 
described keeping a weight with every word/document reference in the 
index and the weight would be adjusted using GA techniques.

Unfortunely, I don't have the reference for the article but, it shouldn't 
be too hard to find.  (The article didn't seem to go into actual 
recall/precision figures so, I have no idea how well it worked other than 
it seemed to help somewhat.)

Anybody else know something about this?

-Art


