On Mon, 16 Jun 2003, Tamas Dombos wrote:
> Has anyone tried truly integrating full text search into the eprints
> software? By integration I mean a common search interface, with both
> metadata AND full text search at once. (All the full text search I found
> was two separate searches and of course the results list didn't use the
> metadata, but some excerpt from the text.)
>
> One (apparently) easy way to do this would be to use an already existing
> search engine (like ht://dig), and pass the results (eprint ID is in the
> URL of the documents) to the eprints software, and combine the id search
> with the metadata search from the original search form. I know that some
> information would be lost (relevence ordering of the full text search,
> for example) but this seems to be a good start. Now this works in theory,
> but I have no idea how difficult it is to implement this.
> http://www.htdig.org/ Any suggestions?
Full-text search capability for Eprint Archives is an excellent
idea, and I am sure it will become an essential feature.
Htdig is already implemented by at least one Eprints.org site:
http://archivesic.ccsd.cnrs.fr/search.html
but it is certainly *highly* desirable to have inverted full-text
for all Eprint Archives.
It needs some thought whether it makes more sense to invert full-text
at each local archive, or at a harvester level (google-like).
Chris?
Stevan Harnad
PS Here are some prior discussions of inverted full text
in the Amsci Forum:
http://makeashorterlink.com/?T24532FE4
and OAI-General:
http://makeashorterlink.com/?N26521FE4
Received on Mon Jun 16 2003 - 00:33:49 BST