Adrien Grand created LUCENE-8510:
------------------------------------
Summary: Better maximum scores for boolean queries
Key: LUCENE-8510
URL: https://issues.apache.org/jira/browse/LUCENE-8510
Project: Lucene - Core
Issue Type: Improvement
Reporter: Adrien Grand
The maximum score of boolean queries is the sum of the maximum score of each
clause. In the common case that each sub query runs on the same field, we might
be able to get better estimates though. Say the query looks like "body:A
body:B": if the maximum score of "body:A" is obtained for a document of length
X and the maximum score of "body:B" is obtained for a document of length Y!=X
then the sum of these maximum scores may never be reached: since both clauses
have to match the same document, the maximum scores of each clause should be
computed for the same value of the norm.
Improving our APIs to improve this sounds challenging, but I was wondering that
we could obtain almost the same result by enabling to sort by norm value, so
that we would combine maximum scores of clauses that have the same norm.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
