[
https://issues.apache.org/jira/browse/LUCENE-7347?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15341156#comment-15341156
]
David Smiley commented on LUCENE-7347:
--------------------------------------
Ah, ok -- thanks for that nice explanation. On a 4.x project recently that had
some high per-field boosts and minShouldMax allowing some terms to not match, I
was finding the high per-field boosts would far exceed the default coord
calculation (which is just a simple ratio, as you know). Aside from reducing
the boosts (which is a balancing act), I toyed with the idea of customizing the
Similarity to give a huge penalty for each non-matching term. I didn't get
around to it but now I wonder how BM25 would fair -- wether the per-field boost
is within the control of what 'k1' saturates, or wether there might be some
other param.
Any way, if you guys really think we're better off without coord then I'm not
standing in your way. Seems a shame to see it go, though.
> Remove queryNorm and coords
> ---------------------------
>
> Key: LUCENE-7347
> URL: https://issues.apache.org/jira/browse/LUCENE-7347
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Assignee: Adrien Grand
>
> These two features are specific to TF-IDF and introduce some complexity (see
> eg. handling of coords in BooleanWeight) and bugs/corner-cases (see eg. how
> taking the query norm into account causes scoring challenges on LUCENE-7337).
> Since we made BM25 the default in 6.0, I propose that we remove these
> TF-IDF-specific features in 7.0.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
