Alessandro Benedetti created SOLR-17226:
-------------------------------------------
Summary: Learning To Rank feature logging should support query
limits
Key: SOLR-17226
URL: https://issues.apache.org/jira/browse/SOLR-17226
Project: Solr
Issue Type: Improvement
Security Level: Public (Default Security Level. Issues are Public)
Components: contrib - LTR
Affects Versions: 9.5.0
Reporter: Alessandro Benedetti
Currently the feature extraction that happens at re-scoring time is separated
from the feature extraction that happens with the feature logger transformer.
This has various implications, especially for caching and query limits support:
-Features extracted for rescoring are not cached and re-used for logging and
vice-versa
- when features are extracted for logging and presentation, no query limit is
checked
Scope of this issue is just to address the second point:
Query Limits should be checked when feature extraction happens for feature
logging.
If a limit is exceeded a partial result should returned (flagged as partial).
A possible approach is to check the query limit after each document.
So the partial result will present the feature vector only for the docs Solr
processed within the time limit and no more.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
