[
https://issues.apache.org/jira/browse/LUCENE-8482?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16607101#comment-16607101
]
Adrien Grand commented on LUCENE-8482:
--------------------------------------
This looks promising! There seems to be an issue with multi-valued fields since
in the 1D case, we can assume that values come in order to more easily find the
closest one to the origin, but in this case, we probably need to look at every
point to find the closest one. This doesn't invalidate your benchmarks though
as they were run on a single-valued field.
I'm wondering what is the costly bit that requires sampling for performance to
be ok. For the record, when updating the iterator in setMinCompetitiveScore, we
don't need to make sure that all documents will have a greater score: it's fine
if some documents still have a score that is less than the minimum competitive
score. Maybe we could try out to only check the bounding box in the points
visitor in case the bottleneck is caused by haversin computations and the
costly distance predicate. It means we will only fail to exclude documents that
are within the bounding box of the circle but not in the circle itself.
> Boosting by geo distance
> ------------------------
>
> Key: LUCENE-8482
> URL: https://issues.apache.org/jira/browse/LUCENE-8482
> Project: Lucene - Core
> Issue Type: New Feature
> Reporter: Adrien Grand
> Priority: Minor
> Attachments: LUCENE-8482.patch
>
>
> Similarly to LUCENE-8340 it would be nice to have an easy and efficient way
> to fold geo distance into scoring formulas in order to boost by proximity.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
