[
https://issues.apache.org/jira/browse/LUCENE-6276?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14980447#comment-14980447
]
Adrien Grand commented on LUCENE-6276:
--------------------------------------
Some suggestions:
- could the match costs be computed eagerly instead of lazily, like we compute
the costs of DocIdSetIterators?
- can you move the utility methods to compute costs of phrases from
TwoPhaseIterator into PhraseWeight/SpanNearQuery. I don't like leaking
implementation details of specific TwoPhaseIterators into TwoPhaseIterator.
- some implementations of matchCost return 0 (eg. RandomAccessWeight); I'm
fine with not implementing every match cost for now, but could you leave a TODO
and use a higher constant (eg. 100)? I think it's safer to assume that such
implementations are costly until they are implemented correctly.
- re: ReqExclScorer: indeed I think we should try to take into account the
cost of advancing the excluded scorer. If this is complicated, I'm fine with
tackling this problem later.
- re: DisjunctionScorer: I think your current definition of the match cost is
fine. Maybe we could improve it by weighting the match cost of each
TwoPhaseIterator by the cost of their approximation. I think it would make
sense since we will call TwoPhaseIterator.matches() more often if their
approximation matches more documents.
- AssertingSpans/AsseringScorer ensure that the match cost is >= 0. Maybe we
should also check for NaN and document acceptable return values in the
documentation of #matchCost?
> Add matchCost() api to TwoPhaseDocIdSetIterator
> -----------------------------------------------
>
> Key: LUCENE-6276
> URL: https://issues.apache.org/jira/browse/LUCENE-6276
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Robert Muir
> Attachments: LUCENE-6276-ExactPhraseOnly.patch,
> LUCENE-6276-NoSpans.patch, LUCENE-6276-NoSpans2.patch, LUCENE-6276.patch,
> LUCENE-6276.patch, LUCENE-6276.patch, LUCENE-6276.patch, LUCENE-6276.patch
>
>
> We could add a method like TwoPhaseDISI.matchCost() defined as something like
> estimate of nanoseconds or similar.
> ConjunctionScorer could use this method to sort its 'twoPhaseIterators' array
> so that cheaper ones are called first. Today it has no idea if one scorer is
> a simple phrase scorer on a short field vs another that might do some geo
> calculation or more expensive stuff.
> PhraseScorers could implement this based on index statistics (e.g.
> totalTermFreq/maxDoc)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
