lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Erick Erickson" <erickerick...@gmail.com>
Subject Re: Calculation of fieldNorm causes irritating effect of sort order
Date Thu, 02 Oct 2008 13:13:44 GMT
Another possibility (and I'm not sure it'll work, but what the heck) would
be
to create a Filter for active ideas. So rather than add a "category:14"
clause,
you create a Category14Filter that you send to the query along with your
+type:idea +alltext:betyg clauses. Now, category won't be considered
in scoring (as I understand it, anyway). See the various Searcher.search()
methods that take a Filter.

Filters are pretty inexpensive to create, so if this works for you I'd just
measure the time it takes to create a filter rather than cache them. But
if you do want to cache filters, CachingWrapperFilter may be your friend.

NOTE: I have no clue whether this will do what you want, but I vaguely
remember from wildcard filter discussions that it might. And it would still
allow boosts on the fields you were searching on....

Best
Erick

On Thu, Oct 2, 2008 at 8:47 AM, Jimi Hullegård <jimi.hullegard@mogul.com>wrote:

> Erik wrote:
> >
> > On Oct 2, 2008, at 7:39 AM, Jimi Hullegård wrote:
> > > Is it possible to disable the lengthNorm calculation for particular
> > > fields?
> >
> > Yes, use Field#setOmitNorms(true) when indexing.
>
> Ok, thanks. I will just have to look on how to do this the best way (since
> the CMS is handling most of the indexing stuff).
>
> But apparently this setOmitNorms(true) also disables boosting aswell. That
> is ok for now, but what if we want to use boosting in the future? Is there
> no way to disable the length normalization while still keeping the boost
> calculation?
>
> /Jimi
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-user-help@lucene.apache.org
>
>

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