[
https://issues.apache.org/jira/browse/LUCENE-10590?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17542068#comment-17542068
]
Michael Sokolov edited comment on LUCENE-10590 at 5/25/22 2:09 PM:
-------------------------------------------------------------------
bq. Does the indexing logic rely on tie breaking by node ID? If not, maybe
index-time graph search could stop as soon as the k-th nearest vector is equal
to the input vector?
Seems like that could work although to date we use the same search
implementation at index time and search time, which is a nice simplification.
Perhaps in such a case we could also sacrifice the docid tiebreaking given that
is going to be best effort only
was (Author: sokolov):
> Does the indexing logic rely on tie breaking by node ID? If not, maybe
> index-time graph search could stop as soon as the k-th nearest vector is
> equal to the input vector?
Seems like that could work although to date we use the same search
implementation at index time and search time, which is a nice simplification.
Perhaps in such a case we could also sacrifice the docid tiebreaking given that
is going to be best effort only
> Indexing all zero vectors leads to heat death of the universe
> -------------------------------------------------------------
>
> Key: LUCENE-10590
> URL: https://issues.apache.org/jira/browse/LUCENE-10590
> Project: Lucene - Core
> Issue Type: Bug
> Reporter: Michael Sokolov
> Priority: Major
>
> By accident while testing something else, I ran a luceneutil test indexing 1M
> 100d vectors where all the vectors were all zeroes. This caused indexing to
> take a very long time (~40x normal - it did eventually complete) and the
> search performance was similarly bad. We should not degrade by orders of
> magnitude with even the worst data though.
> I'm not entirely sure what the issue is, but perhaps as long as we keep
> finding hits that are "better" we keep exploring the graph, where better
> means (score, -docid) >= (lowest score, -docid). If that's right and all docs
> have the same score, then we probably need to either switch to > (but this
> could lead to poorer recall in normal cases) or introduce some kind of
> minimum score threshold?
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
