vigyasharma commented on PR #14009: URL: https://github.com/apache/lucene/pull/14009#issuecomment-2902502742
> Lucene typically rewrites the Query first, and then createWeight on the simplest form. That's a good callout. I'll override the `rewrite` method to rewrite the inner query. That will keep the query cacheable. If inner query rewrite is only called in `createWeight` then each weight can get a different set of knn hits which goes against caching. > we no longer limit the final results by K, so if the inner queries is doing over-sampling, the rerank queries will return the full over-sampled results as well. This gets handled by the `k` you pass to searcher.search(), and the top docs collector you use. It's same as any other query. > would it make sense to accept a DoubleValuesSource (that can compute dot products against a vector field) rather than limiting this to vector dot products only interesting idea, let me see what that would look like. -- 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: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
