diff options
author | Geir Storli <geirst@yahooinc.com> | 2023-08-11 13:16:11 +0000 |
---|---|---|
committer | Geir Storli <geirst@yahooinc.com> | 2023-08-15 13:47:49 +0000 |
commit | 6fbe8e9a17f3bb90f8a8f539ad56308df601ac5b (patch) | |
tree | a4ef9b7f073b3fe91f53bfdb7d8d38cf89375cd8 /container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java | |
parent | 4902b1a4209eb26cfaa22c4527821be89566cc65 (diff) |
Control the auto-adjustment of targetHits in ANN using post-filtering.
When searching the HNSW index in a post-filtering case,
targetHits is auto-adjusted in an effort to still expose targetHits hits to first-phase ranking after post-filtering.
The following formula is now used to ensure an upper bound of adjustedTargetHits,
avoiding that the search in the HNSW index takes too long.
adjustedTargetHits = min(targetHits / estimatedHitRatio, targetHits * targetHitsMaxAdjustmentFactor).
The target-hits-max-adjustment-factor can be set in a rank profile and overriden per query.
The value is in the range [1.0,inf], with the default being 20.0.
When setting this to 1.0, auto-adjustment of targetHits is effectively disabled.
Diffstat (limited to 'container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java')
-rw-r--r-- | container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java b/container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java index 800b3a1ba89..99d6959441a 100644 --- a/container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java +++ b/container-search/src/main/java/com/yahoo/search/query/properties/QueryProperties.java @@ -91,6 +91,7 @@ public class QueryProperties extends Properties { addDualCasedRM(map, Matching.MINHITSPERTHREAD, GetterSetter.of(query -> query.getRanking().getMatching().getMinHitsPerThread(), (query, value) -> query.getRanking().getMatching().setMinHitsPerThread(asInteger(value, 0)))); addDualCasedRM(map, Matching.POST_FILTER_THRESHOLD, GetterSetter.of(query -> query.getRanking().getMatching().getPostFilterThreshold(), (query, value) -> query.getRanking().getMatching().setPostFilterThreshold(asDouble(value, 1.0)))); addDualCasedRM(map, Matching.APPROXIMATE_THRESHOLD, GetterSetter.of(query -> query.getRanking().getMatching().getApproximateThreshold(), (query, value) -> query.getRanking().getMatching().setApproximateThreshold(asDouble(value, 0.05)))); + addDualCasedRM(map, Matching.TARGET_HITS_MAX_ADJUSTMENT_FACTOR, GetterSetter.of(query -> query.getRanking().getMatching().getTargetHitsMaxAdjustmentFactor(), (query, value) -> query.getRanking().getMatching().setTargetHitsMaxAdjustmentFactor(asDouble(value, 20.0)))); map.put(CompoundName.fromComponents(Ranking.RANKING, Ranking.MATCH_PHASE, MatchPhase.ATTRIBUTE), GetterSetter.of(query -> query.getRanking().getMatchPhase().getAttribute(), (query, value) -> query.getRanking().getMatchPhase().setAttribute(asString(value, null)))); map.put(CompoundName.fromComponents(Ranking.RANKING, Ranking.MATCH_PHASE, MatchPhase.ASCENDING), GetterSetter.of(query -> query.getRanking().getMatchPhase().getAscending(), (query, value) -> query.getRanking().getMatchPhase().setAscending(asBoolean(value, false)))); |