lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Raghavendra Prabhu" <>
Subject Re: lucene highlighter
Date Wed, 22 Mar 2006 09:06:38 GMT
Hi Mark

 Currently both of the terms have the same score (weightage)

As you mentioned,  i would want it to be decreased so during the next run
for selecting second fragment, term1 has less weightage and term2 which has
not been selected has more weightage


On 3/22/06, mark harwood <> wrote:
> >>How can i adjust the lucene highlighter to make sure
> >> that atleast each term is displayed in the query
> result
> First some, basic things to sanity check:
> * A classic problem: are you using compatible
> analyzers for tokenizing the query and the document
> content (both index time and highlight time)? Term2
> may not be being produced at all.
> * Are you selecting only one fragment and using a
> fragmenter implementation that means Term1 and Term2
> don't happen to fall within the scope of this single
> fragment?
> If both of these checks turn out OK I suspect what is
> happening is that term2 is weighted significantly less
> than term1 (based on idf and query boosts) and the
> highlighter may be continually selecting multiple
> fragments with term1 in preference to selecting any
> fragments which only contain the lower scoring term2.
> If this is the case and you really want to ensure that
> term2 gets shown then you can use a custom Scorer
> implementation that influences the highlighter
> according to your preferences. Such an implementation
> could, for example, score fragments that are merely
> repetitions of the same "hits" (ie your term1) with a
> decreasing value. This would then allow the fragments
> with term2 to be considered more strongly for
> selection.
> Hope this helps
> Mark
> ___________________________________________________________
> To help you stay safe and secure online, we've developed the all new
> Yahoo! Security Centre.
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message