Details
-
Task
-
Status: Closed
-
Minor
-
Resolution: Fixed
-
None
-
None
-
None
-
New
Description
Computing scores on top hits is fine, but the current way it is implemented - at collection time - requires to read/decode more freqs/norms and compute more scores than necessary. It would be more efficient to compute scores of top hits as a post-collection step by only advancing the scorer to hits that made the top-N list.