iprithv commented on PR #16069: URL: https://github.com/apache/lucene/pull/16069#issuecomment-4640207949
thanks for the suggestion! I tried the filter leading approach..I ran into these two of issues, filter leading path processes all filter docs in the window and only later discards docs that don't match any scoring clause right..with leapfrog though, we only visit docs that are actually in the intersection, so as the intersection ratio gets lower there ends up being more wasted work on filter docs that never contribute to the result. also, once all clauses are treated as optional, we lose the ability to use filterCompetitiveHits between clause applications. here for top-k queries, when minCompetitiveScore starts increasing, MAXSCORE can skip a lot of non competitive docs. filter leading path seems to give up some of that advantage, so there is also some loss of MAXSCORE partitioning. not sure if I'm missing something here though.. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
