Package cc.mallet.topics

Class HierarchicalLDA

java.lang.Object

cc.mallet.topics.HierarchicalLDA

All Implemented Interfaces:

java.io.Serializable

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

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor	Description
HierarchicalLDA()

Method Summary

Modifier and Type	Method	Description
void	calculateNCRP(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights, cc.mallet.topics.HierarchicalLDA.NCRPNode node, double weight)
void	calculateWordLikelihood(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights, cc.mallet.topics.HierarchicalLDA.NCRPNode node, double weight, com.carrotsearch.hppc.IntIntHashMap[] typeCounts, double[] newTopicWeights, int level, int iteration)
double	empiricalLikelihood(int numSamples, InstanceList testing)	For use with empirical likelihood evaluation: sample a path through the tree, then sample a multinomial over topics in that path, then return a weighted sum of words.
void	estimate(int numIterations)
void	initialize(InstanceList instances, InstanceList testing, int numLevels, Randoms random)
static void	main(java.lang.String[] args)	This method is primarily for testing purposes.
void	printNode(cc.mallet.topics.HierarchicalLDA.NCRPNode node, int indent, boolean withWeight)
void	printNodes()
void	printNodes(boolean withWeight)
void	printState()	Writes the current sampling state to the file specified in stateFile.
void	printState(java.io.PrintWriter out)	Write a text file describing the current sampling state.
void	propagateTopicWeight(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights, cc.mallet.topics.HierarchicalLDA.NCRPNode node, double weight)	Propagate a topic weight to a node and all its children.
static HierarchicalLDA	read(java.io.File f)
void	samplePath(int doc, int iteration)
void	sampleTopics(int doc)
void	setAlpha(double alpha)
void	setEta(double eta)
void	setGamma(double gamma)
void	setProgressDisplay(boolean showProgress)	This parameter determines whether the sampler outputs shows progress by outputting a character after every iteration.
void	setStateFile(java.lang.String stateFile)
void	setTopicDisplay(int interval, int words)
void	write(java.io.File serializedModelFile)

Methods inherited from class java.lang.Object

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

Constructor Detail

HierarchicalLDA

public HierarchicalLDA()

Method Detail

setAlpha

public void setAlpha(double alpha)

setGamma

public void setGamma(double gamma)

setEta

public void setEta(double eta)

setStateFile

public void setStateFile(java.lang.String stateFile)

setTopicDisplay

public void setTopicDisplay(int interval,
                            int words)

setProgressDisplay

public void setProgressDisplay(boolean showProgress)

This parameter determines whether the sampler outputs shows progress by outputting a character after every iteration.

initialize

public void initialize(InstanceList instances,
                       InstanceList testing,
                       int numLevels,
                       Randoms random)

estimate

public void estimate(int numIterations)

samplePath

public void samplePath(int doc,
                       int iteration)

calculateNCRP

public void calculateNCRP(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights,
                          cc.mallet.topics.HierarchicalLDA.NCRPNode node,
                          double weight)

calculateWordLikelihood

public void calculateWordLikelihood(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights,
                                    cc.mallet.topics.HierarchicalLDA.NCRPNode node,
                                    double weight,
                                    com.carrotsearch.hppc.IntIntHashMap[] typeCounts,
                                    double[] newTopicWeights,
                                    int level,
                                    int iteration)

propagateTopicWeight

public void propagateTopicWeight(com.carrotsearch.hppc.ObjectDoubleHashMap<cc.mallet.topics.HierarchicalLDA.NCRPNode> nodeWeights,
                                 cc.mallet.topics.HierarchicalLDA.NCRPNode node,
                                 double weight)

Propagate a topic weight to a node and all its children. weight is assumed to be a log.

sampleTopics

public void sampleTopics(int doc)

printState

public void printState()
                throws java.io.IOException,
                       java.io.FileNotFoundException

Writes the current sampling state to the file specified in stateFile.

Throws:

java.io.IOException

java.io.FileNotFoundException

printState

public void printState(java.io.PrintWriter out)
                throws java.io.IOException

Write a text file describing the current sampling state.

Throws:

java.io.IOException

printNodes

public void printNodes()

printNodes

public void printNodes(boolean withWeight)

printNode

public void printNode(cc.mallet.topics.HierarchicalLDA.NCRPNode node,
                      int indent,
                      boolean withWeight)

empiricalLikelihood

public double empiricalLikelihood(int numSamples,
                                  InstanceList testing)

For use with empirical likelihood evaluation: sample a path through the tree, then sample a multinomial over topics in that path, then return a weighted sum of words.

write

public void write(java.io.File serializedModelFile)

read

public static HierarchicalLDA read(java.io.File f)
                            throws java.lang.Exception

Throws:

java.lang.Exception

main

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

This method is primarily for testing purposes. The HierarchicalLDATUI class has a more flexible interface for command-line use.