[
https://issues.apache.org/jira/browse/LUCENE-9335?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17316059#comment-17316059
]
Adrien Grand commented on LUCENE-9335:
--------------------------------------
[~zacharymorn] Yes that would be one idea. In the BMM paper
(http://engineering.nyu.edu/~suel/papers/bmm.pdf) BMM is usually a bit slower
than BMW but not always. I'd be curious to know whether we observe the same
result in Lucene.
Since we introduced BMW there have been a few reports that top-level
disjunctions got slower. This is usually because there are many clauses in a
disjunction that have about the same max score and BMW can hardly skip
evaluating documents. In such cases we pay for the BMW overhead without
enjoying any benefits. Because BMM has less overhead, I would expect it to
perform better in these worst-case scenarios, so I wonder if we should look
into using BMM for top-level disjunctions in general.
> Add a bulk scorer for disjunctions that does dynamic pruning
> ------------------------------------------------------------
>
> Key: LUCENE-9335
> URL: https://issues.apache.org/jira/browse/LUCENE-9335
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Priority: Minor
>
> Lucene often gets benchmarked against other engines, e.g. against Tantivy and
> PISA at [https://tantivy-search.github.io/bench/] or against research
> prototypes in Table 1 of
> [https://cs.uwaterloo.ca/~jimmylin/publications/Grand_etal_ECIR2020_preprint.pdf].
> Given that top-level disjunctions of term queries are commonly used for
> benchmarking, it would be nice to optimize this case a bit more, I suspect
> that we could make fewer per-document decisions by implementing a BulkScorer
> instead of a Scorer.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
