As a part of Project Ockham, I have started integrating Find More Like
This One functionality into one of my bibliographic indexes. See:
http://tinyurl.com/c5htp
The link above will search an index of about 250,000 OAI records for
the term "hemogloban". It should find zero hits because the word is
spelled incorrectly, but the interface will suggest a number of
alternative spellings. Once these spellings are used, the system will
respond with synonyms as well as a suggested abbreviation, Hb. The
system is not perfect; right now it only really works with singe-word
queries.
I am using using Net::OAI::Harvester to collect the OAI data. MyLibrary
is used to manage the harvested content and create reports. I use
swish-e to index/search the data. Spelling corrections employ Aspell,
and the synonyms are generated through the use of WordNet.
Fun!
--
Eric Lease Morgan
University Libraries of Notre Dame
|