[Koha-devel] Searching: Zebra

[Joshua Ferraro]

Hi everyone,

I've spent the last couple of days playing around with Indexdata's
Zebra (http://www.indexdata.dk/zebra/). Zebra is another example
of a textual database engine.

Zebra can natively index MARC records, so I started by dumping out
NPL's records (which took about 20 minutes).

Indexing 150K records took just under 3 minutes and all the searches
I've done using Zebra's integrated Z39.50 Server have taken less
than a second. I haven't had a chance to write a wrapper yet to
integrate the search into our existing code but you can try it out
at the LOC's generic Z39.50 gateway:

http://www.loc.gov/cgi-bin/zgate?ACTION=INIT&FORM_HOST_PORT=/prod/www/data/z3950/locils.html,66.213.78.76,9999&CI=034920

This link will allow you to search NPL's 150K database (you could
also use any Z39.50 client (like Yaz) to do the search. The server's
IP is 66.213.78.76:9999 and the db name is VOYAGER (because that's 
the only way I could get the LOC's gateway to connect).

Zebra is very high performance (it's in use for systems with tens
of millions of records). It supports incremental database updates
on live systems, exact boolean search expressions, relevance-ranked
free-text queries (not sure about stemming).

Of the searching ideas we've come up with so far this one is by far
the most impressive. There are still some other textual database engines
out there (lucene, clucene, xapian, namzu, to name a few), so if anyone
would like to try these out and commit some sample indexing/searching
scripts I'd be happy to try them out.

Coments, questions?
-- 
Joshua Ferraro               VENDOR SERVICES FOR OPEN-SOURCE SOFTWARE
President, Technology       migration, training, maintenance, support
LibLime                  Koha ILS, Mambo Intranet, DiscrimiNet Filter
address@hidden |  Full Demos at http://liblime.com  |  1(888)KohaILS