dragon.ir.classification

Class NBClassifier

java.lang.Object

dragon.ir.classification.AbstractClassifier

dragon.ir.classification.NBClassifier

All Implemented Interfaces:

Classifier

Direct Known Subclasses:

NigamActiveLearning, SemanticNBClassifier

public class NBClassifier
extends AbstractClassifier

Naive Bayesian classifier which uses Laplacian smoothing

Copyright: Copyright (c) 2005

Company: IST, Drexel University

Version:

1.0

Author:

Davis Zhou

Field Summary

Fields

Modifier and Type	Field and Description
protected DoubleVector	classPrior
protected DoubleVector	lastClassProb
protected DoubleFlatDenseMatrix	model

Fields inherited from class dragon.ir.classification.AbstractClassifier

arrLabel, classNum, doctermMatrix, featureSelector, indexReader, validatingDocSet

Constructor Summary

Constructors

Constructor and Description
NBClassifier(IndexReader indexReader)
NBClassifier(SparseMatrix doctermMatrix)
NBClassifier(String modelFile)

Method Summary

Modifier and Type	Method and Description
int	classify(IRDoc doc) Classify one particular document
int	classify(Row doc) Classify one particular document
protected DoubleVector	getClassPrior(DocClassSet docSet)
int[]	rank() Rank all class labels.
void	saveModel(String modelFile) Save the trained classifier model to a file which can be used to restore the classifier later.
void	train(DocClassSet trainingDocSet) This method trains the classifier with the training document set.

Methods inherited from class dragon.ir.classification.AbstractClassifier

classify, classify, classify, getClassLabel, getDocTermMatrix, getFeatureSelector, getIndexReader, getRow, setFeatureSelector, train, trainFeatureSelector

Methods inherited from class java.lang.Object

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

Field Detail

model

protected DoubleFlatDenseMatrix model

classPrior

protected DoubleVector classPrior

lastClassProb

protected DoubleVector lastClassProb

Constructor Detail

NBClassifier

public NBClassifier(String modelFile)

NBClassifier

public NBClassifier(IndexReader indexReader)

NBClassifier

public NBClassifier(SparseMatrix doctermMatrix)

Method Detail

train

public void train(DocClassSet trainingDocSet)

Description copied from interface: Classifier

This method trains the classifier with the training document set.

Parameters:

trainingDocSet - training document set

getClassPrior

protected DoubleVector getClassPrior(DocClassSet docSet)

classify

public int classify(IRDoc doc)

Description copied from interface: Classifier

Classify one particular document

Specified by:

classify in interface Classifier

Overrides:

classify in class AbstractClassifier

Parameters:

doc - the index of the document is stored in the IRDoc object

Returns:

the index of the category of this document. The index starts from zero.

classify

public int classify(Row doc)

Description copied from interface: Classifier

Classify one particular document

Parameters:

doc - document represented by a Row object

Returns:

the index of the category of this document. The index starts from zero.

rank

public int[] rank()

Description copied from interface: Classifier

Rank all class labels. The classify(Row doc) method should be called before calling this method.

Returns:

the ranking of class labels. The first and last element of the returned array contain the most possible label and the least possible label, respectively.

saveModel

public void saveModel(String modelFile)

Description copied from interface: Classifier

Save the trained classifier model to a file which can be used to restore the classifier later. If the model is not trained yet, this method does nothing. The classifier restored from this model file later can only execute the method classify(Row doc) to determine the label of a document.

Parameters:

modelFile - output file name