Re: [EP-underground] Full text search

Sounds like a reinvention of google. 

I would rather use OAI to supply the URL's of the full texts and
allow third parties to extract information in their own manner.

Currently it would stall the system horribly to have it index the 
entire paper as it's done when the paper is "commited" to the db -
which is after EVERY page of the submission form.

I'm planning on making this happen seperately with a script in a 
later version, at which point we could add a multiple, longtext, 
non-user-editable field to eprints, each value of which was all
the extracted text from the document.

However such text extraction systems can be ropey. I'm in no
hurry to implement it in the main disribution, but am happy to
add "hooks" for people to add it themselves.

On Tue, Jun 18, 2002 at 09:42:04PM +0100, Silvio Macedo wrote:
> 
> Wouldn't it be possible to do text extraction of the article's text
> (it is possible for word, pdf and ps) and allow full text** search
> capabilities, through something like MNGOsearch (inverse word indexing) ?
> 
> The answer I know, is that it is.. but, have you considered it ?
> Does NEC citeseer, www.researchindex.com do it ?
> This way, metadata would be secondary...
> 
> ** possibly limited to year ranges, or subject...
> 
> Silvio
> 
> 
> On Tue, 18 Jun 2002, ePrints Support wrote:
> > EPrints2 can support other datasets than unqualified dublin core (oai_dc)
> > but doing so is probably more effort than many people can be bothered with.
> 
> -- 
> ````````` Silvio Emanuel Nunes Barbosa de Macedo (PhD Student) '''''''''''
> [email protected]                                         [email protected]
> Intelligent and Interactive Systems,EEE-Pg         Telecom. and Multimedia
> Imperial College, University of London                         INESC Porto
> Exhibition Road,                                       Pc da Republica, 93
> London SW7 2BT, England                            4050-497 Porto Portugal
> Tel:+44 20 75946323                                    Tel:+351  222094220

-- 

 Christopher Gutteridge                   [email protected]
 ePrints2 Coder, Support and Stuff        +44 23 8059 4833