[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13610835#comment-13610835 ] Commit Tag Bot commented on LUCENE-4208: [branch_4x commit] David Wayne Smiley http://svn.apache.org/viewvc?view=revisionrevision=1385122 LUCENE-4208 makeQuery return ConstantScoreQuery, standardize makeDistanceValueSource behavior Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Assignee: David Smiley Fix For: 4.0 Attachments: LUCENE-4208_makeQuery_return_ConstantScoreQuery_and_remake_TwoDoublesStrategy.patch, LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch, LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13455768#comment-13455768 ] Chris Male commented on LUCENE-4208: Things are looking pretty good, we're almost there. - Where are we on multi-valued fields? In the documentation on makeDistanceValueSource it doesn't say what happens when multiple values are indexed. Do we support that in the ValueSource implementations? is the behaviour undefined? If it is supposed to be defined, can we document it? - Returns a ValueSource useful as a score Can we drop this claim? Part of the reason we've moved to having ConstantScoreQuerys is that it isn't clear what the score for the queries should be. This value isn't useful for every spatial operation or implementation. Once these have gotten addressed, I'm +1 for committing. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 Attachments: LUCENE-4208_makeQuery_return_ConstantScoreQuery_and_remake_TwoDoublesStrategy.patch, LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch, LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[
https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13455788#comment-13455788
]
David Smiley commented on LUCENE-4208:
--
Here are my proposed updates to the javadocs:
makeDistanceValueSource:
{code}
/**
* Make a ValueSource returning the distance between the center of the
* indexed shape and {@code queryPoint}. If there are multiple indexed shapes
* then the closest one is chosen.
*/
{code}
makeRecipDistanceValueSource:
{code}
/**
* Returns a ValueSource with values ranging from 1 to 0, depending inversely
* to the distance from {@link
#makeDistanceValueSource(com.spatial4j.core.shape.Point)}.
* The formula is codec/(d + c)/code where 'd' is the distance and 'c' is
* one tenth the distance to the farthest edge from the center. Thus the
* scores will be 1 for indexed points at the center of the query shape and as
* low as ~0.1 at its furthest edges.
*/
public final ValueSource makeRecipDistanceValueSource(Shape queryShape) {
{code}
Spatial distance relevancy should use score of 1/distance
-
Key: LUCENE-4208
URL: https://issues.apache.org/jira/browse/LUCENE-4208
Project: Lucene - Core
Issue Type: New Feature
Components: modules/spatial
Reporter: David Smiley
Fix For: 4.0
Attachments:
LUCENE-4208_makeQuery_return_ConstantScoreQuery_and_remake_TwoDoublesStrategy.patch,
LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch,
LUCENE-4208_makeQuery_return_ConstantScoreQuery,_standardize_makeDistanceValueSource_behav.patch
The SpatialStrategy.makeQuery() at the moment uses the distance as the score
(although some strategies -- TwoDoubles if I recall might not do anything
which would be a bug). The distance is a poor value to use as the score
because the score should be related to relevancy, and the distance itself is
inversely related to that. A score of 1/distance would be nice. Another
alternative is earthCircumference/2 - distance, although I like 1/distance
better. Maybe use a different constant than 1.
Credit: this is Chris Male's idea.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13453746#comment-13453746 ] Chris Male commented on LUCENE-4208: I disagree that makeQuery shouldn't exist. There are optimizations to be had in Query code, such as using BooleanQuery and its associated highly optimized scorer algs. I think it should continue to exist but should have a default implementation that creates a CSQ by calling makeFilter. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13454581#comment-13454581 ] Chris Male commented on LUCENE-4208: bq. TwoDoubles is getting overhauled to support the dateline and any query shape--should probably go into another issue. Yes please! Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 Attachments: LUCENE-4208_makeQuery_return_ConstantScoreQuery_and_remake_TwoDoublesStrategy.patch The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13453692#comment-13453692 ] David Smiley commented on LUCENE-4208: -- This being said, something relatively simple should be done to advance this issue for v4 (and I'll do the work now). It comes down to what is the score of makeQuery()? The strategies are doing different things right now. Here are 3 options I can think of: # Standardize on the reciprocal distance using some specific formula like c/(d+c) where c is 1/10th the distance to the nearest query shape edge. # Documented loosely to allow any reasonable shape similarity so that it will at least sort sensibly (this precludes returning the distance directly). # Add a ValueSource parameter to makeQuery(). This allows the caller to pick something. If null is supplied then you get a ConstantScoreQuery. A convenience method on Strategy can be provided to return a ValueSource of the reciprocal distance formula if the caller wants to use that. The #3 the ValueSource parameter seems appealing to me but then I remember why makeQuery() exists instead of letting the user combine the distance valueSource makeFilter. If the filter requires the valueSource (such as a [working] TwoDoubles given a query shape other than a rect), then the valueSource might be invoked twice: first for matching and second for the score. One idea to alter option #3 is to make the argument an implementation of a single-method interface that accepts the distance valueSource and returns a valueSource for the score. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13453708#comment-13453708 ] Chris Male commented on LUCENE-4208: I don't think there is a clear solution here. But I feel ValueSource provides maximum flexibility going forward. If we continue to support makeValueSource then people can sort, or include it in their query if they want, or just retrieve the value at some later stage. makeQuery() should just return a ConstantScoreQuery. We can consider in future versions what if anything we want to do around its score. WRT to TwoDoubles. This Strategy was a nice start to this work awhile back and was designed to replicate existing point-distance functionality. But it has huge limitations and it constantly feels like we're being held back by it. Every Strategy has its limitations, and I dont feel we should hold back changes just because it impacts TwoDoubles. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13453743#comment-13453743 ] David Smiley commented on LUCENE-4208: -- If we standardize on ConstantScoreQuery then it's dubious if makeQuery() should even exist -- just use makeFilter(). If the user code wants a Query then it can wrap the Filter. It could first detect if the Filter is a QueryWrapperFilter and if so grab the Query object from that. This algorithm could even be the default implementation in SpatialStrategy. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13451508#comment-13451508 ] Chris Male commented on LUCENE-4208: I actually totally agree with David here. Using ValueSource (instead of my SpatialSimilarity idea) is an excellent solution which leverages existing Lucene code. Having it this way means that even if a Strategy has a custom Query implementation (maybe for performance reasons) it would still be possible to make use of the ValueSource in scoring. I definitely think we should expose this on a per Strategy basis rather than all Strategys as some Strategys may not be able to compute distance and we shouldn't force them to. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13451430#comment-13451430 ] Itamar Syn-Hershko commented on LUCENE-4208: What's the status of this? are query results being properly sorted based on distance? Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[
https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13451462#comment-13451462
]
David Smiley commented on LUCENE-4208:
--
I have some in-progress code I need to finish, then post as a patch.
I think the most important thing is for the strategies to expose the distance
as a ValueSource. This is the case currently with PrefixTreeStrategy and
TwoDoublesStrategy. With this, you can do what you want -- inverse reciprocal
distance or whatever. By default, I'd like a Query's score to be something
like that and this is what this issue is about, but even without it you could
do something like:
{code}
new FilteredQuery(new FunctionQuery(myValueSource),
spatialStrategy.makeFilter(...))
{code}
Spatial distance relevancy should use score of 1/distance
-
Key: LUCENE-4208
URL: https://issues.apache.org/jira/browse/LUCENE-4208
Project: Lucene - Core
Issue Type: New Feature
Components: modules/spatial
Reporter: David Smiley
Fix For: 4.0
The SpatialStrategy.makeQuery() at the moment uses the distance as the score
(although some strategies -- TwoDoubles if I recall might not do anything
which would be a bug). The distance is a poor value to use as the score
because the score should be related to relevancy, and the distance itself is
inversely related to that. A score of 1/distance would be nice. Another
alternative is earthCircumference/2 - distance, although I like 1/distance
better. Maybe use a different constant than 1.
Credit: this is Chris Male's idea.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[
https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13424358#comment-13424358
]
Chris Male commented on LUCENE-4208:
Having thought about this more I think the best way forward is to just emulate
free-text queries and have a {{SpatialSimilarity}} abstraction. I'm not sure
of the exact nature of the API for this but I think there are times with using
1/x is sufficient and there are probably times when a more convoluted algorithm
fits. We should allow the consumer to control what they choose.
I think the Similarity should be given the Query Shape, the matched docID and
the current SpatialOperation as a minimum. I'd like to somehow see a way to
also pass in a pre-computed distance (for Queries that compute it as part of
their matching) and possibly the matched grid hash for anything using the
PrefixTrees. We might have to have subclasses for those, or maybe a Command or
something, I'm not sure.
Other benefits:
- We immediately open up the ability to have more complex similarity scores
based on overlap percentage or anything really.
- It is plausible that a SpatialSimilarity might use a cache of indexed Shapes
to facilitate more complex algorithms. By having this abstraction we offload
the caching from the main API.
- It is also plausible that a SpatialSimilarity instance could be misused to
cache calculated distances if the consumer so wanted.
I think we should consider whether we want SpatialSimilarities to also be given
the current IndexReader (and so be able to use it in any caches or other
lookups) or whether we want them to be IR independent.
We will also need some custom Queries to actually make use of the
SpatialSimilarity. Need to think this one through a little.
Spatial distance relevancy should use score of 1/distance
-
Key: LUCENE-4208
URL: https://issues.apache.org/jira/browse/LUCENE-4208
Project: Lucene - Core
Issue Type: New Feature
Components: modules/spatial
Reporter: David Smiley
Fix For: 4.0
The SpatialStrategy.makeQuery() at the moment uses the distance as the score
(although some strategies -- TwoDoubles if I recall might not do anything
which would be a bug). The distance is a poor value to use as the score
because the score should be related to relevancy, and the distance itself is
inversely related to that. A score of 1/distance would be nice. Another
alternative is earthCircumference/2 - distance, although I like 1/distance
better. Maybe use a different constant than 1.
Credit: this is Chris Male's idea.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13423851#comment-13423851 ] Chris Male commented on LUCENE-4208: I'm considered this obfuscates the actual distance too much, making it difficult to retrieve x again. It's not impossible but suddenly anybody wanting to retrieve the actual distance must calculate c again. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-4208) Spatial distance relevancy should use score of 1/distance
[ https://issues.apache.org/jira/browse/LUCENE-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13423638#comment-13423638 ] David Smiley commented on LUCENE-4208: -- Thinking more about this, I like the overall curve shape of the inverted distance but it definitely has problems. For example if the query center is practically on top of some data point then the score ends up being close to infinity. And how quickly the score curve slides is doesn't vary based on the overall query shape; I think it should. The reciprocal makes sense: c/(x+c) where x = the distance. 'c' is roughly 1/10th of the distance to the query shape edge. Spatial distance relevancy should use score of 1/distance - Key: LUCENE-4208 URL: https://issues.apache.org/jira/browse/LUCENE-4208 Project: Lucene - Core Issue Type: New Feature Components: modules/spatial Reporter: David Smiley Fix For: 4.0 The SpatialStrategy.makeQuery() at the moment uses the distance as the score (although some strategies -- TwoDoubles if I recall might not do anything which would be a bug). The distance is a poor value to use as the score because the score should be related to relevancy, and the distance itself is inversely related to that. A score of 1/distance would be nice. Another alternative is earthCircumference/2 - distance, although I like 1/distance better. Maybe use a different constant than 1. Credit: this is Chris Male's idea. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
