Package cc.mallet.cluster.neighbor_evaluator

Class PairwiseEvaluator

java.lang.Object

cc.mallet.cluster.neighbor_evaluator.ClassifyingNeighborEvaluator

cc.mallet.cluster.neighbor_evaluator.PairwiseEvaluator

All Implemented Interfaces:

NeighborEvaluator, java.io.Serializable

public class PairwiseEvaluator
extends ClassifyingNeighborEvaluator

Uses a Classifier over pairs of Instances to score Neighbor. Currently only supports AgglomerativeNeighbors.

Since:

1.0

Version:

1.0

Author:

"Aron Culotta"

See Also:

ClassifyingNeighborEvaluator, Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	PairwiseEvaluator.Average
static interface	PairwiseEvaluator.CombiningStrategy	Specifies how to combine a set of pairwise scores into a cluster-wise score.
static class	PairwiseEvaluator.Maximum
static class	PairwiseEvaluator.Minimum

Constructor Summary

Constructors

Constructor	Description
PairwiseEvaluator(Classifier classifier, java.lang.String scoringLabel, PairwiseEvaluator.CombiningStrategy combiningStrategy, boolean mergeFirst)

Method Summary

Modifier and Type	Method	Description
double	evaluate(Neighbor neighbor)
double[]	evaluate(Neighbor[] neighbors)
void	reset()	Reset the state of the evaluator.
java.lang.String	toString()

Methods inherited from class cc.mallet.cluster.neighbor_evaluator.ClassifyingNeighborEvaluator

getClassifier

Methods inherited from class java.lang.Object

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

Constructor Detail

PairwiseEvaluator

public PairwiseEvaluator(Classifier classifier,
                         java.lang.String scoringLabel,
                         PairwiseEvaluator.CombiningStrategy combiningStrategy,
                         boolean mergeFirst)

Parameters:

classifier - Classifier to assign scores to Neighbors for which a pair of Instances has been merged.

scoringLabel - The predicted label that corresponds to a positive example (e.g. "YES").

combiningStrategy - How to combine the pairwise scores (e.g. max, mean, ...).

mergeFirst - If true, score all edges involved in a merge. If false, only score the edges that cross the boundaries of the clusters being merged.

Method Detail

evaluate

public double[] evaluate(Neighbor[] neighbors)

Specified by:

evaluate in interface NeighborEvaluator

Overrides:

evaluate in class ClassifyingNeighborEvaluator

Returns:

One score per neighbor. A higher score indicates that the modified Clustering is preferred.

evaluate

public double evaluate(Neighbor neighbor)

Specified by:

evaluate in interface NeighborEvaluator

Overrides:

evaluate in class ClassifyingNeighborEvaluator

Returns:

A higher score indicates that the modified Clustering is preferred.

reset

public void reset()

Description copied from interface: NeighborEvaluator

Reset the state of the evaluator.

Specified by:

reset in interface NeighborEvaluator

Overrides:

reset in class ClassifyingNeighborEvaluator

toString

public java.lang.String toString()

Overrides:

toString in class ClassifyingNeighborEvaluator