lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sebastian Marius Kirsch <skir...@sebastian-kirsch.org>
Subject Re: Lucene + LSI
Date Tue, 13 Dec 2005 11:26:15 GMT
On Tue, Dec 13, 2005 at 10:53:42AM +0000, adasal wrote:
> There seem to be quite a few alternatives around. I would be interested in
> comments on the following:-
> The work at NITLE <http://www.nitle.org/tools/semantic/search.htm>
> using Contextual
> Network Search (CNS) a graph-based alternative to LSI.

Hi,

perhaps I can say a few words about CNS.

I evaluated spreading activation search (which is what CNS is based
on) for my master's thesis. I must conclude that SA is a rather fickle
method -- if you read Preece's thesis, you will see that it can be
used to implement a number of retrieval algorithms and techniques. The
devil is in the details: SA depends strongly on a suitable set of
adjustments and constraints which express exactly the kind of
retrieval (or inference) algorithm you want. 

For applications of SA in Retrieval, see for example the following
publications:

http://www.sebastian-kirsch.org/moebius/reading.html#Crestani1997a
http://www.sebastian-kirsch.org/moebius/reading.html#Crestani2000
http://www.sebastian-kirsch.org/moebius/reading.html#Belew1989
http://www.sebastian-kirsch.org/moebius/reading.html#Pirolli1996

You will notice that Crestani stresses the fact that no successful
commercial system based on SA was ever produced. I am also unaware of
any peer-reviewed publications on CNS, or follow-up publications by
NITLE. Maciej Ceglowski has since left NITLE, as far as I know.


So if I was you, I'd place my bet on probabilistic or linear algebra
methods like pLSI/LSI. 

If anyone wants to chime in, please feel free to comment ;)

Regards, Sebastian

-- 
Sebastian Kirsch <skirsch@sebastian-kirsch.org> [http://www.sebastian-kirsch.org/]

NOTE: New email address! Please update your address book.

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message