Hi,
Has anyone experimented information theory based expanded query
weight boosting for Lucene?
When the user query is small, there are several ways to expand
the query terms by synonym terms, morph terms etc. I read several
articles on how different boosting levels affect the precision and
recall of the expanded query. In the context of Lucene, is there a
systematic study conducted to analyze different boosting schemes? It
would be nice to have a theoretical proof rather than empirically
setting boost levels for the expanded query terms. The main disadvantage
of empirical boost levels is that there is no cross system comparison
and is highly dependent on the test bed.
Thanks,
Rajesh Munavalli
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
