[
https://issues.apache.org/jira/browse/LUCENE-8013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16219682#comment-16219682
]
Robert Muir commented on LUCENE-8013:
-------------------------------------
+1, the test would be simple.
at the same time maybe we should reconsider this method: it is strange the
phrase "fuzziness" is allowed here, but term fuzziness is not for example. We
could just use {{1/(N+1)}} in SloppyPhraseQuery and that's the end of it.
BooleanSimilarity will still score the same way, since it returns 1 for
score(adjustedFreq), so we don't have evidence for any use-case for an
alternate implementation?
I feel similar about scorePayload. If you want to score payloads, you must pass
PayloadFunction logic to the payload-based query *and* write a custom
similarity. So you have to split your custom logic across two places and the
apis seem awkward because of that. So maybe additionally the scorePayload
should be moved to PayloadFunction api so that its all in one place? Seems like
we can potentially make it much better that way.
> Test that SimScorer.computeSlopFactor doesn't increase as the distance goes up
> ------------------------------------------------------------------------------
>
> Key: LUCENE-8013
> URL: https://issues.apache.org/jira/browse/LUCENE-8013
> Project: Lucene - Core
> Issue Type: Task
> Reporter: Adrien Grand
> Priority: Minor
> Fix For: master (8.0)
>
>
> We should ensure that computeSlopFactor is always (strictly) positive, equal
> to 1 when distance is 0, and doesn't increase when the distance goes up.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
