lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chuck Williams" <>
Subject RE: Boolean Scorer
Date Fri, 10 Dec 2004 21:16:45 GMT
I'm just a user and haven't tried Paul's DisjunctionScorer yet, but I
must say this sounds like a great idea.  The ability to specialize
combineScores() is a significant advantage.  I'd suggest adding
DisjunctionMaxScorer in addition to DisjunctionSumScorer and
DisjunctionSumCoordScorer in the default distribution, and adding a
variant to MultiFieldQueryParser that uses these to expand the query
differently (expand the innermost clauses to search the multiple fields
with Max rather than duplicating the original query across each field; I
have code to do that for my MaxDisjunctionQuery which could be trivially
changed to use DisjunctionMaxScorer if this was desired).  I expect the
variant MultiFieldQueryParser would be what most users want most of the
time (e.g., it does a much better job searching for "red elephants"
across fields "title" and "description").


  > -----Original Message-----
  > From: Doug Cutting []
  > Sent: Friday, December 10, 2004 12:36 PM
  > To: Lucene Developers List
  > Subject: Re: Boolean Scorer
  > Christoph Goller wrote:
  > > I think we should change BooleanScorer. An easy way would be to
  > the
  > > bucket
  > > list before it is used. Do you think that would affect performance
  > > dramatically?
  > I think it would make it slower.
  > > Otherwise we should reimplement BooleanScorer. I haven't looked
  > the
  > > DisjunctionScorer patch in Bugzilla yet. Maybe it's a good
  > point.
  > I think we should incorporate Paul's code into CVS.  This algorithm
  > be slower in some cases, but it may also be faster in some cases.
  > should add a static method to switch back to the old implementation,
  > encourage folks to benchmark their code.  If it proves no slower
then we
  > could remove the old implementation altogether.
  > What do others think?
  > Paul's code is in:
  > Has anyone tried this?
  > Doug
  > To unsubscribe, e-mail:
  > For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message