dragon.ir.clustering.clustermodel

Class MultinomialClusterModel

java.lang.Object

dragon.ir.clustering.clustermodel.AbstractClusterModel

dragon.ir.clustering.clustermodel.MultinomialClusterModel

All Implemented Interfaces:

ClusterModel

public class MultinomialClusterModel
extends AbstractClusterModel

Multinormial model based clustering

Copyright: Copyright (c) 2005

Company: IST, Drexel University

Version:

1.0

Author:

Davis Zhou

Field Summary

Fields inherited from class dragon.ir.clustering.clustermodel.AbstractClusterModel

clusterNum, featureFilter

Constructor Summary

Constructors

Constructor and Description
MultinomialClusterModel(int clusterNum, IndexReader indexReader)
MultinomialClusterModel(int clusterNum, IndexReader indexReader, double bkgCoefficient)
MultinomialClusterModel(int clusterNum, IndexReader indexReader, IndexReader topicIndexReader, DoubleSparseMatrix topicTransMatrix, double transCoefficient, double bkgCoefficient)
MultinomialClusterModel(int clusterNum, IndexReader indexReader, IndexReader topicIndexReader, KnowledgeBase kngBase, double transCoefficient, double bkgCoefficient)

Method Summary

Modifier and Type	Method and Description
double	getBackgroundCoefficient()
double	getDistance(IRDoc doc, int clusterID)
double	getTranslationCoefficient()
void	setBackgroundCoefficient(double bkgCoefficient)
void	setClusterNum(int clusterNum)
void	setDocCluster(DocCluster cluster) Iterative partional approaches (e.g.
void	setFeatureFilter(FeatureFilter featureFilter) A feature selector is set.
void	setTranslationCoefficient(double transCoefficient)

Methods inherited from class dragon.ir.clustering.clustermodel.AbstractClusterModel

getClusterNum, getDistance, getFeatureFilter, setDocClusters

Methods inherited from class java.lang.Object

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

Constructor Detail

MultinomialClusterModel

public MultinomialClusterModel(int clusterNum,
                               IndexReader indexReader)

MultinomialClusterModel

public MultinomialClusterModel(int clusterNum,
                               IndexReader indexReader,
                               double bkgCoefficient)

MultinomialClusterModel

public MultinomialClusterModel(int clusterNum,
                               IndexReader indexReader,
                               IndexReader topicIndexReader,
                               DoubleSparseMatrix topicTransMatrix,
                               double transCoefficient,
                               double bkgCoefficient)

MultinomialClusterModel

public MultinomialClusterModel(int clusterNum,
                               IndexReader indexReader,
                               IndexReader topicIndexReader,
                               KnowledgeBase kngBase,
                               double transCoefficient,
                               double bkgCoefficient)

Method Detail

getTranslationCoefficient

public double getTranslationCoefficient()

setTranslationCoefficient

public void setTranslationCoefficient(double transCoefficient)

getBackgroundCoefficient

public double getBackgroundCoefficient()

setBackgroundCoefficient

public void setBackgroundCoefficient(double bkgCoefficient)

setFeatureFilter

public void setFeatureFilter(FeatureFilter featureFilter)

Description copied from interface: ClusterModel

A feature selector is set. After that, the excluded features do not count in cluster model any more.

Specified by:

setFeatureFilter in interface ClusterModel

Overrides:

setFeatureFilter in class AbstractClusterModel

getDistance

public double getDistance(IRDoc doc,
                          int clusterID)

setClusterNum

public void setClusterNum(int clusterNum)

Parameters:

clusterNum - the number of clusters

setDocCluster

public void setDocCluster(DocCluster cluster)

Description copied from interface: ClusterModel

Iterative partional approaches (e.g. K-Means) need to re-compute the cetroid of each cluster after each iteration. This method gives the chance to re-compute the centroid.

Parameters:

cluster - the new cluster