[
https://issues.apache.org/jira/browse/LUCENE-4933?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13631266#comment-13631266
]
Hoss Man commented on LUCENE-4933:
----------------------------------
bq. As far as I can tell, its had this bug all along.
definitely a non-intentional fuck up from the very begining
bq. patch removing score(int)/tf(int) so the trap doesn't even exist anymore.
i only skimmed the patch and didn't understand most of it (need to look at in
context more) but if i'm understanding the crux of it you don't just mean from
SSS, you mean you want to remove tf(int) from TFIDFSimilarity entirely?
isn't that kind of a baby vs bathwater situation?
historically the value of having both tf(int) and tf(float) was that people
could choose to implement alternative functions for dealing with phrase
frequency (using tf(float)) vs single term query's (using tf(int)) ... is that
still possible in some other way in all of the other changes in your patch that
i didn't see in my quick skim?
> SweetSpotSimilarity doesnt override tf(float)
> ---------------------------------------------
>
> Key: LUCENE-4933
> URL: https://issues.apache.org/jira/browse/LUCENE-4933
> Project: Lucene - Core
> Issue Type: Bug
> Components: core/query/scoring
> Affects Versions: 2.0.0
> Reporter: Robert Muir
> Attachments: LUCENE-4933.patch
>
>
> This means its scoring is not really right: it only applies to term queries
> and exact phrase queries, but not e.g. sloppy phrase queries and spans.
> As far as I can tell, its had this bug all along.
--
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: [email protected]
For additional commands, e-mail: [email protected]
