Class SeededKnnVectorQuery
See "Lexically-Accelerated Dense Retrieval" (Kulkarni, Hrishikesh and MacAvaney, Sean and Goharian, Nazli and Frieder, Ophir). In SIGIR '23: Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval Pages 152 - 162
- WARNING: This API is experimental and might change in incompatible ways in the next release.
-
Field Summary
FieldsModifier and TypeFieldDescriptionprotected final Stringprotected final Queryprotected final intprotected final KnnSearchStrategy -
Method Summary
Modifier and TypeMethodDescriptionprotected TopDocsapproximateSearch(LeafReaderContext context, Bits acceptDocs, int visitedLimit, KnnCollectorManager knnCollectorManager) booleanOverride and implement query instance equivalence properly in a subclass.protected TopDocsexactSearch(LeafReaderContext context, DocIdSetIterator acceptIterator, QueryTimeout queryTimeout) static SeededKnnVectorQueryfromByteQuery(KnnByteVectorQuery knnQuery, Query seed) Construct a new SeededKnnVectorQuery instance for a byte vector fieldstatic SeededKnnVectorQueryfromFloatQuery(KnnFloatVectorQuery knnQuery, Query seed) Construct a new SeededKnnVectorQuery instance for a float vector fieldgetField()intgetK()protected KnnCollectorManagergetKnnCollectorManager(int k, IndexSearcher searcher) inthashCode()Override and implement query hash code properly in a subclass.protected TopDocsmergeLeafResults(TopDocs[] perLeafResults) Merges all segment-level kNN results to get the index-level kNN results.rewrite(IndexSearcher indexSearcher) Expert: called to re-write queries into primitive queries.Prints a query to a string, withfieldassumed to be the default field and omitted.voidvisit(QueryVisitor visitor) Recurse through the query tree, visiting any child queries.Methods inherited from class org.apache.lucene.search.Query
classHash, createWeight, sameClassAs, toString
-
Field Details
-
field
-
k
protected final int k -
filter
-
searchStrategy
-
-
Method Details
-
fromFloatQuery
Construct a new SeededKnnVectorQuery instance for a float vector field- Parameters:
knnQuery- the knn query to be seededseed- a query seed to initiate the vector format search- Returns:
- a new SeededKnnVectorQuery instance
- WARNING: This API is experimental and might change in incompatible ways in the next release.
-
fromByteQuery
Construct a new SeededKnnVectorQuery instance for a byte vector field- Parameters:
knnQuery- the knn query to be seededseed- a query seed to initiate the vector format search- Returns:
- a new SeededKnnVectorQuery instance
- WARNING: This API is experimental and might change in incompatible ways in the next release.
-
toString
Description copied from class:QueryPrints a query to a string, withfieldassumed to be the default field and omitted. -
rewrite
Description copied from class:QueryExpert: called to re-write queries into primitive queries. For example, a PrefixQuery will be rewritten into a BooleanQuery that consists of TermQuerys.Callers are expected to call
rewritemultiple times if necessary, until the rewritten query is the same as the original query.The rewrite process may be able to make use of IndexSearcher's executor and be executed in parallel if the executor is provided.
- Throws:
IOException- See Also:
-
approximateSearch
protected TopDocs approximateSearch(LeafReaderContext context, Bits acceptDocs, int visitedLimit, KnnCollectorManager knnCollectorManager) throws IOException - Throws:
IOException
-
getKnnCollectorManager
-
exactSearch
protected TopDocs exactSearch(LeafReaderContext context, DocIdSetIterator acceptIterator, QueryTimeout queryTimeout) throws IOException - Throws:
IOException
-
mergeLeafResults
Merges all segment-level kNN results to get the index-level kNN results.The default implementation delegates to
TopDocs.merge(int, TopDocs[])to find the overall topk, which requires input results to be sorted.This method is useful for reading and / or modifying the final results as needed.
- Parameters:
perLeafResults- array of segment-level kNN results.- Returns:
- index-level kNN results (no constraint on their ordering).
-
visit
Description copied from class:QueryRecurse through the query tree, visiting any child queries.- Parameters:
visitor- a QueryVisitor to be called by each query in the tree
-
equals
Description copied from class:QueryOverride and implement query instance equivalence properly in a subclass. This is required so thatQueryCacheworks properly.Typically a query will be equal to another only if it's an instance of the same class and its document-filtering properties are identical to those of the other instance. Utility methods are provided for certain repetitive code.
- See Also:
-
hashCode
public int hashCode()Description copied from class:QueryOverride and implement query hash code properly in a subclass. This is required so thatQueryCacheworks properly.- See Also:
-
getField
- Returns:
- the knn vector field where the knn vector search happens.
-
getK
public int getK()- Returns:
- the max number of results the KnnVector search returns.
-
getFilter
- Returns:
- the filter that is executed before the KnnVector search happens. Only the results accepted by this filter are returned by the KnnVector search.
-