[tlug] search for fulltext-searchengine

["Stephen J. Turnbull" <stephen@example.com>]

Christian Horn writes:

 > - provide a knowledge-database via webinterface to users

Don't know about this part.

 > - provide a search-function that indexes 
 >    - the knowledge-database contents
 >    - and office-documents, pdf, textfiles in a directory

For indexing, have you looked at Xapian?  I haven't actually worked
with it myself, but Roundup uses it for full-text searches of the
database.  But I think it only does a few varieties of text (eg, TeX)
out-of-the-box.  It does have good Python bindings, which means
writing a front-end to the indexer is not too hard.

 > - be able to do all this with kanji

Not sure about this for Xapian.

 > namazu.org looks nice for a fulltext-searchengine.

Namazu is a high-maintenance girlfriend.  Give her one rock, she
expects another one weekly.  We (xemacs.org) gave up on it.

I will be looking at Xapian more carefully in the not-to-distant
future.

A suprising candidate is FreeWAIS.  I know one guy who swears by it
because of its ability to do proximity search, which turns out to do
very well despite being far more primitive than Xapian's probabilistic
search technology.