edu.stanford.nlp.sentiment

Class SentimentModel

java.lang.Object

edu.stanford.nlp.sentiment.SentimentModel

All Implemented Interfaces:

java.io.Serializable

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

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
TwoDimensionalMap<java.lang.String,java.lang.String,SimpleMatrix>	binaryClassification CxN+1, where N = size of word vectors, C is the number of classes
int	binaryClassificationSize How many elements a classification matrix has
TwoDimensionalMap<java.lang.String,java.lang.String,SimpleTensor>	binaryTensors 2Nx2NxN, where N is the size of the word vectors
int	binaryTensorSize How many elements the binary transformation tensors have
TwoDimensionalMap<java.lang.String,java.lang.String,SimpleMatrix>	binaryTransform Nx2N+1, where N is the size of the word vectors
int	binaryTransformSize How many elements a transformation matrix has
int	numBinaryMatrices Cached here for easy calculation of the model size; TwoDimensionalMap does not return that in O(1) time
int	numClasses How many classes the RNN is built to test against
int	numHid Dimension of hidden layers, size of word vectors, etc
int	numUnaryMatrices Cached here for easy calculation of the model size; TwoDimensionalMap does not return that in O(1) time
java.util.Map<java.lang.String,SimpleMatrix>	unaryClassification CxN+1, where N = size of word vectors, C is the number of classes
int	unaryClassificationSize How many elements a classification matrix has
java.util.Map<java.lang.String,SimpleMatrix>	wordVectors

Constructor Summary

Constructors

Constructor and Description
SentimentModel(RNNOptions op, java.util.List<Tree> trainingTrees) The traditional way of initializing an empty model suitable for training.

Method Summary

Methods

Modifier and Type	Method and Description
java.lang.String	basicCategory(java.lang.String category)
SimpleMatrix	getBinaryClassification(java.lang.String left, java.lang.String right)
SimpleTensor	getBinaryTensor(java.lang.String left, java.lang.String right)
SimpleMatrix	getBinaryTransform(java.lang.String left, java.lang.String right)
SimpleMatrix	getClassWForNode(Tree node)
SimpleTensor	getTensorForNode(Tree node)
SimpleMatrix	getUnaryClassification(java.lang.String category)
java.lang.String	getVocabWord(java.lang.String word)
SimpleMatrix	getWForNode(Tree node)
SimpleMatrix	getWordVector(java.lang.String word)
static SentimentModel	loadSerialized(java.lang.String path)
double[]	paramsToVector()
void	printParamInformation(int index)
void	saveSerialized(java.lang.String path)
int	totalParamSize()
void	vectorToParams(double[] theta)

Methods inherited from class java.lang.Object

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

Field Detail

binaryTransform

public TwoDimensionalMap<java.lang.String,java.lang.String,SimpleMatrix> binaryTransform

Nx2N+1, where N is the size of the word vectors

binaryTensors

public TwoDimensionalMap<java.lang.String,java.lang.String,SimpleTensor> binaryTensors

2Nx2NxN, where N is the size of the word vectors

binaryClassification

public TwoDimensionalMap<java.lang.String,java.lang.String,SimpleMatrix> binaryClassification

CxN+1, where N = size of word vectors, C is the number of classes

unaryClassification

public java.util.Map<java.lang.String,SimpleMatrix> unaryClassification

CxN+1, where N = size of word vectors, C is the number of classes

wordVectors

public java.util.Map<java.lang.String,SimpleMatrix> wordVectors

numClasses

public final int numClasses

How many classes the RNN is built to test against

numHid

public final int numHid

Dimension of hidden layers, size of word vectors, etc

numBinaryMatrices

public final int numBinaryMatrices

Cached here for easy calculation of the model size; TwoDimensionalMap does not return that in O(1) time

binaryTransformSize

public final int binaryTransformSize

How many elements a transformation matrix has

binaryTensorSize

public final int binaryTensorSize

How many elements the binary transformation tensors have

binaryClassificationSize

public final int binaryClassificationSize

How many elements a classification matrix has

numUnaryMatrices

public final int numUnaryMatrices

Cached here for easy calculation of the model size; TwoDimensionalMap does not return that in O(1) time

unaryClassificationSize

public final int unaryClassificationSize

How many elements a classification matrix has

Constructor Detail

SentimentModel

public SentimentModel(RNNOptions op,
              java.util.List<Tree> trainingTrees)

The traditional way of initializing an empty model suitable for training.

Method Detail

totalParamSize

public int totalParamSize()

paramsToVector

public double[] paramsToVector()

vectorToParams

public void vectorToParams(double[] theta)

getWForNode

public SimpleMatrix getWForNode(Tree node)

getTensorForNode

public SimpleTensor getTensorForNode(Tree node)

getClassWForNode

public SimpleMatrix getClassWForNode(Tree node)

getWordVector

public SimpleMatrix getWordVector(java.lang.String word)

getVocabWord

public java.lang.String getVocabWord(java.lang.String word)

basicCategory

public java.lang.String basicCategory(java.lang.String category)

getUnaryClassification

public SimpleMatrix getUnaryClassification(java.lang.String category)

getBinaryClassification

public SimpleMatrix getBinaryClassification(java.lang.String left,
                                   java.lang.String right)

getBinaryTransform

public SimpleMatrix getBinaryTransform(java.lang.String left,
                              java.lang.String right)

getBinaryTensor

public SimpleTensor getBinaryTensor(java.lang.String left,
                           java.lang.String right)

saveSerialized

public void saveSerialized(java.lang.String path)

loadSerialized

public static SentimentModel loadSerialized(java.lang.String path)

printParamInformation

public void printParamInformation(int index)