cc.mallet.topics
Class NPTopicModel

java.lang.Object
  cc.mallet.topics.NPTopicModel

All Implemented Interfaces:

java.io.Serializable

public class NPTopicModel
extends java.lang.Object
implements java.io.Serializable

A non-parametric topic model that uses the "minimal path" assumption to reduce bookkeeping.

Author:

David Mimno

See Also:

Serialized Form

Field Summary

protected double	alpha
protected Alphabet	alphabet
protected double	beta
protected double	betaSum
protected java.util.ArrayList<TopicAssignment>	data
static double	DEFAULT_BETA
protected gnu.trove.TIntIntHashMap	docsPerTopic
protected java.text.NumberFormat	formatter
protected double	gamma
protected int	maxTopic
protected int	numTopics
protected int	numTypes
protected boolean	printLogLikelihood
protected Randoms	random
int	showTopicsInterval
protected gnu.trove.TIntIntHashMap	tokensPerTopic
protected LabelAlphabet	topicAlphabet
protected int	totalDocTopics
protected gnu.trove.TIntIntHashMap[]	typeTopicCounts
int	wordsPerTopic

Constructor Summary

NPTopicModel(double alpha, double gamma, double beta)

Method Summary

void	addInstances(InstanceList training, int initialTopics)
static void	main(java.lang.String[] args)
void	printState(java.io.File f)
void	printState(java.io.PrintStream out)
void	sample(int iterations)
protected void	sampleTopicsForOneDoc(FeatureSequence tokenSequence, FeatureSequence topicSequence)
void	setRandomSeed(int seed)
void	setTopicDisplay(int interval, int n)
java.lang.String	topWords(int numWords)

Methods inherited from class java.lang.Object

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

Field Detail

data

protected java.util.ArrayList<TopicAssignment> data

alphabet

protected Alphabet alphabet

topicAlphabet

protected LabelAlphabet topicAlphabet

maxTopic

protected int maxTopic

numTopics

protected int numTopics

numTypes

protected int numTypes

alpha

protected double alpha

gamma

protected double gamma

beta

protected double beta

betaSum

protected double betaSum

DEFAULT_BETA

public static final double DEFAULT_BETA

See Also:

Constant Field Values

typeTopicCounts

protected gnu.trove.TIntIntHashMap[] typeTopicCounts

tokensPerTopic

protected gnu.trove.TIntIntHashMap tokensPerTopic

docsPerTopic

protected gnu.trove.TIntIntHashMap docsPerTopic

totalDocTopics

protected int totalDocTopics

showTopicsInterval

public int showTopicsInterval

wordsPerTopic

public int wordsPerTopic

random

protected Randoms random

formatter

protected java.text.NumberFormat formatter

printLogLikelihood

protected boolean printLogLikelihood

Constructor Detail

NPTopicModel

public NPTopicModel(double alpha,
                    double gamma,
                    double beta)

Parameters:

alpha - this parameter balances the local document topic counts with the global distribution over topics.

gamma - this parameter is the weight on a completely new, never-before-seen topic in the global distribution.

beta - this parameter controls the variability of the topic-word distributions

Method Detail

setTopicDisplay

public void setTopicDisplay(int interval,
                            int n)

setRandomSeed

public void setRandomSeed(int seed)

addInstances

public void addInstances(InstanceList training,
                         int initialTopics)

sample

public void sample(int iterations)
            throws java.io.IOException

Throws:

java.io.IOException

sampleTopicsForOneDoc

protected void sampleTopicsForOneDoc(FeatureSequence tokenSequence,
                                     FeatureSequence topicSequence)

topWords

public java.lang.String topWords(int numWords)

printState

public void printState(java.io.File f)
                throws java.io.IOException

Throws:

java.io.IOException

printState

public void printState(java.io.PrintStream out)

main

public static void main(java.lang.String[] args)
                 throws java.io.IOException

Throws:

java.io.IOException