lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrzej Bialecki ...@getopt.org>
Subject Re: Pre-filtering for expensive query
Date Thu, 04 Sep 2008 08:20:05 GMT
Grant Ingersoll wrote:
> 
> On Aug 30, 2008, at 3:14 PM, Andrzej Bialecki wrote:

>>
>> I think you can use a FilteredQuery in a BooleanClause. This may be 
>> faster than the filtering code in the Searcher, because the evaluation 
>> is done during scoring and not afterwards. FilteredQuery internally makes
> 
> 
> FYI, not sure if this is exactly what you are talking about Andrzej, but 
> IndexSearcher no longer filters after scoring.  This was changed in 
> https://issues.apache.org/jira/browse/LUCENE-584

Ah, indeed - I was working with 2.3.0 release ... then there should be 
no visible performance difference if using the trunk version of 
IndexSearcher.

The only difference now between the IndexSearcher method and 
ConjunctionScorer would be when the supplied filter would match many 
documents. IndexSearcher always runs skipTo on the filter first, so 
potentially it would stop at many docIds that aren't matching in the 
scorer - whereas the ConjunctionScorer tries to order sub-scorers so 
that "sparse" scorers are checked first.

-- 
Best regards,
Andrzej Bialecki     <><
  ___. ___ ___ ___ _ _   __________________________________
[__ || __|__/|__||\/|  Information Retrieval, Semantic Web
___|||__||  \|  ||  |  Embedded Unix, System Integration
http://www.sigram.com  Contact: info at sigram dot com


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message