lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "patrick o'leary" <pj...@pjaol.com>
Subject Re: LocalLucene/Lucene Spatial
Date Mon, 20 Apr 2009 14:18:31 GMT
Honestly I'm more focused on intelligent ways to do faster and more complex
GIS features.
As I said the most time consuming part is the DistainceFilter, which is
required to sort by distance.
I'm playing with several ideas on how to do those better, and get a win
there.

However if anyone wants to jump in and try other improvements feel free.

On Mon, Apr 20, 2009 at 4:59 AM, Uwe Schindler <uwe@thetaphi.de> wrote:

> Have you thought about subclassing MultiTermQuery and provide a
> FilteredTermEnum? When you do this, the query can be either BooleanQuery or
> a Filter.
>
> -----
> Uwe Schindler
> H.-H.-Meier-Allee 63, D-28213 Bremen
> http://www.thetaphi.de
> eMail: uwe@thetaphi.de
>
> > -----Original Message-----
> > From: patrick o'leary [mailto:pjaol@pjaol.com]
> > Sent: Monday, April 20, 2009 3:33 AM
> > To: java-user@lucene.apache.org
> > Subject: Re: LocalLucene/Lucene Spatial
> >
> > 1) The only reason for ConstantScoreQuery is because it lets me convert
> > the
> > LocalLucene filter to a query.
> >
> > 2) TierRangeQuery- not at the moment, we're not really doing a range
> > search,
> > we're pulling predefined id's (e.g. a shape) from the index using the
> term
> > enumerator.
> >
> > Average time take for a bounding box shape lookup is in the order of 3 to
> > 4
> > ms, repeating that with a lat / long bounding box, and your looking at
> > 100+
> > ms. Been there, done that.
> >
> > As for bitsets, I'm looking at openbitset, but the getDocSetId is causing
> > me
> > some problems with a QueryWrapperFilter, where I'm intersecting the text
> > search with the bounding shape
> > filter pre-distance filter..
> >
> > The reality is the heaviest part is the DistanceFilter- not the bounding
> > box
> > part.
> >
> > On Sun, Apr 19, 2009 at 8:06 PM, John Wang <john.wang@gmail.com> wrote:
> >
> > > Is there a reason the Query build is from a bitset via a
> > ConstantScoreQuery
> > > instead a RangeQuery? Seems we would be paying a penalty for loading
> the
> > > bitset, esp the bitset would be rather sparse.
> > >
> > > Furthermore, is TrieRangeQuery planning to be somehow used in the
> > spatial
> > > package?
> > >
> > > Thanks
> > >
> > > -John
> > >
> > > On Wed, Mar 18, 2009 at 11:10 AM, Grant Ingersoll <gsingers@apache.org
> > > >wrote:
> > >
> > > > In the trunk, see contrib/spatial.  I believe it is all committed and
> > > being
> > > > worked on by interested contributors:
> > > > http://svn.apache.org/repos/asf/lucene/java/trunk/contrib/spatial/
> > > >
> > > > -Grant
> > > >
> > > >
> > > > On Mar 18, 2009, at 12:05 PM, Jamie Johnson wrote:
> > > >
> > > >  I am working on a project that is already using Lucene (through
> > > Hibernate
> > > >> Search) to perform full text queries and have since come across
> > several
> > > >> sites with information about LocalLucene/Lucene Spatial.  I was
> > > wondering
> > > >> if
> > > >> there was a timeline for having this work implemented that anyone
> was
> > > >> aware
> > > >> of?
> > > >>
> > > >
> > > >
> > > > ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> > > > For additional commands, e-mail: java-user-help@lucene.apache.org
> > > >
> > > >
> > >
>
>
> ---------------------------------------------------------------------
> 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