Abstract searching via Paracite

From: Mike Jewell <moj_at_ecs.soton.ac.uk>
Date: Thu, 6 Nov 2003 17:06:19 +0000

The Paracite article locating tool now supports searching for and
identifying documents via abstracts, as well as the existing functionality
for location by citation. This new interface is available from the
Paracite front page:

http://paracite.eprints.org

The abstract search uses a common word discovery technique as a very
simple (yet very effective) approach to identifying documents, with an
article being reduced to 10 keywords after eliminating the most common
English words. These keywords can be stored as a persistent
representation of the document, or passed into other search engines
(initial tests with Google were very successful, with a search on the 10
keywords frequently returning the full article as the top result).

The Paracite interface offers a direct link to the Google search once an
abstract has been processed, but also uses a database of preparsed
documents to provide possible immediate matches. This database currently
contains a large portion of the archives based at the University of
Southampton, and this will be expanded to provide wider coverage of
existing open repositories.

The next release of the ParaTools Perl modules will include the code for
the generation of the keywords from documents, and the web service will
also be updated to allow for searching by keyword signature.

 Mike Jewell <mike_at_mikesroom.org> | Leftfield - Open Up
     http://www.mikesroom.org |
                                   |
  http://www.soundclick.com/darko | Hi Jimbo. Dennis. Really appreciate the
    http://paracite.eprints.org | help on the income tax. You wanna help on
     http://www.beautifier.org | the audit now? -- "The Rockford Files"
Received on Thu Nov 06 2003 - 17:06:19 GMT

This archive was generated by hypermail 2.3.0 : Fri Dec 10 2010 - 19:47:08 GMT