[
https://issues.apache.org/jira/browse/LUCENE-8681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16764163#comment-16764163
]
Mike Sokolov commented on LUCENE-8681:
--------------------------------------
I hope I'm not reading this the right way (?!? :), but I do agree this is
potentially inexact. The point of the math above is that we can bound the
inexactness. The current API assumes the user is OK with some inexact counting,
and this just builds on that. I can see how we might want to provide more
explicit control, but whether or not we do that is a somewhat independent
question.
The assumption this patch makes is that documents are distributed among
segments in a uniform random way. As I think about it, perhaps there are cases
where there would be a correlation - eg if the field is a timestamp, recent
documents could very well be grouped together in a single segment. Although
multi-threaded indexing and merging will tend to spread them around, there
could still be a correlation.
> Prorated early termination
> --------------------------
>
> Key: LUCENE-8681
> URL: https://issues.apache.org/jira/browse/LUCENE-8681
> Project: Lucene - Core
> Issue Type: Improvement
> Components: core/search
> Reporter: Mike Sokolov
> Priority: Major
>
> In this issue we'll exploit the distribution of top K documents among
> segments to extract performance gains when using early termination. The basic
> idea is we do not need to collect K documents from every segment and then
> merge. Rather we can collect a number of documents that is proportional to
> the segment's size plus an error bound derived from the combinatorics seen as
> a (multinomial) probability distribution.
> https://github.com/apache/lucene-solr/pull/564 has the proposed change.
> [~rcmuir] pointed out on the mailing list that this patch confounds two
> settings: (1) whether to collect all hits, ensuring correct hit counts, and
> (2) whether to guarantee that the top K hits are precisely the top K.
> The current patch treats this as the same thing. It takes the position that
> if the user says it's OK to have approximate counts, then it's also OK to
> introduce some small chance of ranking error; occasionally some of the top K
> we return may draw from the top K + epsilon.
> Instead we could provide some additional knobs to the user. Currently the
> public API is {{TopFieldCOllector.create(Sort, int, FieldDoc, int
> threshold)}}. The threshold parameter controls when to apply early
> termination; it allows the collector to terminate once the given number of
> documents have been collected.
> Instead of using the same threshold to control leaf-level early termination,
> we could provide an additional leaf-level parameter. For example, this could
> be a scale factor on the error bound, eg a number of standard deviations to
> apply. The patch uses 3, but a much more conservative bound would be 4 or
> even 5. With these values, some speedup would still result, but with a much
> lower level of ranking errors. A value of MAX_INT would ensure no leaf-level
> termination would ever occur.
> We could also hide the precise numerical bound and offer users a three-way
> enum (EXACT, APPROXIMATE_COUNT, APPROXIMATE_RANK) that controls whether to
> apply this optimization, using some predetermined error bound.
> I posted the patch without any user-level tuning since I think the user has
> already indicated a preference for speed over precision by specifying a
> finite (global) threshold, but if we want to provide finer control, these two
> options seem to make the most sense to me. Providing access to the number of
> standard deviation to allow from the expected distribution gives the user the
> finest control, but it could be hard to explain its proper use.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
