Package it.unimi.di.big.mg4j.search.score

Class BM25FScorer.Visitor

java.lang.Object

it.unimi.di.big.mg4j.search.visitor.AbstractDocumentIteratorVisitor

it.unimi.di.big.mg4j.search.score.BM25FScorer.Visitor

All Implemented Interfaces:

DocumentIteratorVisitor<Boolean>

Enclosing class:

BM25FScorer

protected static final class BM25FScorer.Visitor
extends AbstractDocumentIteratorVisitor

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	expectedIDF	Whether we are running Boldi's variant.
protected double[]	idfPart	Precomputed IDF part.
protected double[]	index2B	For each index, the associated parameter b.
protected double[]	index2BDividedByAvgDocumentSize	Cached value.
protected int	numberOfIndices	The length of TermCollectionVisitor.indices() cached.
protected int	numberOfOffsetsSeen	The number of valid entries in seenOffsetsList.
protected int	numberOfTermsSeen	The number of terms (irrespective of the index) seen up to now and accumulated in seenTermIdsList.
protected int[]	offset2Index	An array indexed by offsets mapping each offset to the corresponding index number.
protected int[]	offset2TermId	For each offset, the corresponding term id in the query.
protected boolean[]	seen	Whether we have already seen a specific term/index pair.
protected int[]	seenOffsetsList	An array accumulating the indices in seen that have been set to true, so to accelerate reset(long).
protected int[]	seenTermIdsList	The list of numberOfTermsSeen term ids that we have seen so far.
protected IntBigList[]	sizes	For each index, its list of document sizes.
protected double[]	sizeWeightComponent	Cached from BM25Scorer.
protected int[]	termId2VirtualCounter	For each term id, its virtual-counter index (to be used to access virtualCount and virtualIdfCount).
protected double[]	virtualCount	An array indexed by term ids used by BM25FScorer.score() to compute virtual counts.
protected double[]	virtualIdfCount	For expected IDF runs, an array indexed by term ids used by BM25FScorer.score() to compute virtual counts combined with IDF scoring.
protected double[]	weight	For each index, its weight.

Constructor Summary

Constructors

Constructor	Description
Visitor(boolean expectedIDF, double[] idfPart, int[] offset2TermId, int[] offset2Index, double[] weight, double[] index2B, IntBigList[] sizes, double[] index2BDividedByAvgDocumentSize)

Method Summary

Modifier and Type	Method	Description
void	reset(long document)
Boolean	visit(IndexIterator indexIterator)	Visits an IndexIterator leaf.

Methods inherited from class it.unimi.di.big.mg4j.search.visitor.AbstractDocumentIteratorVisitor

newArray, prepare, visit, visit, visit, visitPost, visitPre

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

sizeWeightComponent

protected final double[] sizeWeightComponent

Cached from BM25Scorer.

numberOfIndices

protected final int numberOfIndices

The length of TermCollectionVisitor.indices() cached.

seenOffsetsList

protected final int[] seenOffsetsList

An array accumulating the indices in seen that have been set to true, so to accelerate reset(long).

numberOfOffsetsSeen

protected int numberOfOffsetsSeen

The number of valid entries in seenOffsetsList.

seen

protected final boolean[] seen

Whether we have already seen a specific term/index pair.

numberOfTermsSeen

protected int numberOfTermsSeen

The number of terms (irrespective of the index) seen up to now and accumulated in seenTermIdsList.

seenTermIdsList

protected final int[] seenTermIdsList

The list of numberOfTermsSeen term ids that we have seen so far.

termId2VirtualCounter

protected final int[] termId2VirtualCounter

For each term id, its virtual-counter index (to be used to access virtualCount and virtualIdfCount).

sizes

protected final IntBigList[] sizes

For each index, its list of document sizes.

weight

protected final double[] weight

For each index, its weight.

index2B

protected final double[] index2B

For each index, the associated parameter b.

expectedIDF

protected final boolean expectedIDF

Whether we are running Boldi's variant.

offset2TermId

protected final int[] offset2TermId

For each offset, the corresponding term id in the query.

offset2Index

protected final int[] offset2Index

An array indexed by offsets mapping each offset to the corresponding index number.

virtualCount

protected final double[] virtualCount

An array indexed by term ids used by BM25FScorer.score() to compute virtual counts.

virtualIdfCount

protected final double[] virtualIdfCount

For expected IDF runs, an array indexed by term ids used by BM25FScorer.score() to compute virtual counts combined with IDF scoring.

index2BDividedByAvgDocumentSize

protected final double[] index2BDividedByAvgDocumentSize

Cached value.

idfPart

protected final double[] idfPart

Precomputed IDF part.

Constructor Detail

Visitor

public Visitor(boolean expectedIDF,
               double[] idfPart,
               int[] offset2TermId,
               int[] offset2Index,
               double[] weight,
               double[] index2B,
               IntBigList[] sizes,
               double[] index2BDividedByAvgDocumentSize)

Method Detail

visit

public Boolean visit(IndexIterator indexIterator)
              throws IOException

Description copied from interface: DocumentIteratorVisitor

Visits an IndexIterator leaf.

Parameters:

indexIterator - the leaf to be visited.

Returns:

an appropriate return value if the visit should continue, or null.

Throws:

IOException

reset

public void reset(long document)