Package org.apache.lucene.search

Class LateInteractionFloatValuesSource

java.lang.Object

org.apache.lucene.search.DoubleValuesSource

org.apache.lucene.search.LateInteractionFloatValuesSource

All Implemented Interfaces:

SegmentCacheable

public class LateInteractionFloatValuesSource
extends DoubleValuesSource

A DoubleValuesSource that scores documents using similarity between a multi-vector query, and indexed document multi-vectors.

This is useful re-ranking query results using late interaction models, where documents and queries are represented as multi-vectors of composing token vectors. Document vectors are indexed using LateInteractionField.

WARNING: This API is experimental and might change in incompatible ways in the next release.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	LateInteractionFloatValuesSource.ScoreFunction	Defines the function to compute similarity score between query and document multi-vectors

Field Summary

Fields inherited from class org.apache.lucene.search.DoubleValuesSource

SCORES

Constructor Summary

Constructors

Constructor	Description
LateInteractionFloatValuesSource(String fieldName, float[][] queryVector)
LateInteractionFloatValuesSource(String fieldName, float[][] queryVector, VectorSimilarityFunction vectorSimilarityFunction)
LateInteractionFloatValuesSource(String fieldName, float[][] queryVector, VectorSimilarityFunction vectorSimilarityFunction, MultiVectorSimilarity scoreFunction)

Method Summary

Modifier and Type	Method	Description
boolean	equals(Object obj)
DoubleValues	getValues(LeafReaderContext ctx, DoubleValues scores)	Returns a DoubleValues instance for the passed-in LeafReaderContext and scores
int	hashCode()
boolean	isCacheable(LeafReaderContext ctx)
boolean	needsScores()	Return true if document scores are needed to calculate values
DoubleValuesSource	rewrite(IndexSearcher reader)	Return a DoubleValuesSource specialised for the given IndexSearcher
String	toString()

Methods inherited from class org.apache.lucene.search.DoubleValuesSource

constant, explain, fromDoubleField, fromField, fromFloatField, fromIntField, fromLongField, fromQuery, fromScorer, getSortField, similarityToQueryVector, similarityToQueryVector, toLongValuesSource, toSortableLongDoubleValuesSource

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Details

LateInteractionFloatValuesSource

public LateInteractionFloatValuesSource(String fieldName,
 float[][] queryVector)

LateInteractionFloatValuesSource

public LateInteractionFloatValuesSource(String fieldName,
 float[][] queryVector,
 VectorSimilarityFunction vectorSimilarityFunction)

LateInteractionFloatValuesSource

public LateInteractionFloatValuesSource(String fieldName,
 float[][] queryVector,
 VectorSimilarityFunction vectorSimilarityFunction,
 MultiVectorSimilarity scoreFunction)

Method Details

getValues

public DoubleValues getValues(LeafReaderContext ctx,
 DoubleValues scores)
                       throws IOException

Description copied from class: DoubleValuesSource

Returns a DoubleValues instance for the passed-in LeafReaderContext and scores

If scores are not needed to calculate the values (ie returns false, callers may safely pass null for the scores parameter.

Specified by:

getValues in class DoubleValuesSource

Throws:

IOException

needsScores

public boolean needsScores()

Description copied from class: DoubleValuesSource

Return true if document scores are needed to calculate values

Specified by:

needsScores in class DoubleValuesSource

rewrite

public DoubleValuesSource rewrite(IndexSearcher reader)
                           throws IOException

Description copied from class: DoubleValuesSource

Return a DoubleValuesSource specialised for the given IndexSearcher

Implementations should assume that this will only be called once. IndexReader-independent implementations can just return this

Queries that use DoubleValuesSource objects should call rewrite() during Query.createWeight(IndexSearcher, ScoreMode, float) rather than during Query.rewrite(IndexSearcher) to avoid IndexReader reference leakage.

For the same reason, implementations that cache references to the IndexSearcher should return a new object from this method.

Specified by:

rewrite in class DoubleValuesSource

Throws:

IOException

hashCode

public int hashCode()

Specified by:

hashCode in class DoubleValuesSource

equals

public boolean equals(Object obj)

Specified by:

equals in class DoubleValuesSource

toString

public String toString()

Specified by:

toString in class DoubleValuesSource

isCacheable

public boolean isCacheable(LeafReaderContext ctx)

Returns:

true if the object can be cached against a given leaf