edu.umn.cs.nlp.mt.chiang2007
Class LoglinearTranslationFeatures

java.lang.Object
  edu.umn.cs.nlp.mt.chiang2007.LoglinearTranslationFeatures

public class LoglinearTranslationFeatures
extends Object

Represents a set of feature values in a log-linear model, along with a shared set of lambda parameter values.

Version:

$LastChangedDate: 2007-08-28 10:02:47 -0500 (Tue, 28 Aug 2007) $

Author:

Lane Schwartz

Field Summary

static LoglinearTranslationFeatures	DEFAULT_FEATURES
static LoglinearTranslationFeatures	GLUE_RULE_FEATURES

Constructor Summary

LoglinearTranslationFeatures(double sourceGivenTargetLexProbWeight, double targetGivenSourceLexProbWeight, double sourceGivenTargetPhraseProbWeight, double targetGivenSourcePhraseProbWeight, double numberOfWords)

LoglinearTranslationFeatures(LoglinearTranslationFeatures l1, LoglinearTranslationFeatures l2)
Construct a new set of feature values from existing feature values and new weights.

Method Summary

double	getGlueScore()
double	getLanguageModelProb()
double	getLexLogProbSourceGivenTarget()
double	getLexLogProbTargetGivenSource()
int	getNumberOfWords()
double	getPhrasePenalty()
double	getRuleLogProbSourceGivenTarget()
double	getRuleLogProbTargetGivenSource()
double	getScore() Return a score.
static void	setLambdas(double lambda_sourceGivenTargetLexProb, double lambda_targetGivenSourceLexProb, double lambda_sourceGivenTargetPhraseProb, double lambda_targetGivenSourcePhraseProb, double lambda_languageModel, double lambda_wordPenalty, double lambda_phrasePenalty, double lambda_glue)
void	setLanguageModelProb(double languageModelProb) Set the languageModelProb
double[]	toArray() Return the set of feature values as an array.
String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_FEATURES

public static final LoglinearTranslationFeatures DEFAULT_FEATURES

GLUE_RULE_FEATURES

public static final LoglinearTranslationFeatures GLUE_RULE_FEATURES

Constructor Detail

LoglinearTranslationFeatures

public LoglinearTranslationFeatures(double sourceGivenTargetLexProbWeight,
                                    double targetGivenSourceLexProbWeight,
                                    double sourceGivenTargetPhraseProbWeight,
                                    double targetGivenSourcePhraseProbWeight,
                                    double numberOfWords)

LoglinearTranslationFeatures

public LoglinearTranslationFeatures(LoglinearTranslationFeatures l1,
                                    LoglinearTranslationFeatures l2)

Construct a new set of feature values from existing feature values and new weights.

Parameters:

l - Existing feature values

sourceGivenTargetLexProbWeight -

targetGivenSourceLexProbWeight -

sourceGivenTargetPhraseProbWeight -

targetGivenSourcePhraseProbWeight -

numberOfWords -

Method Detail

getGlueScore

public double getGlueScore()

setLambdas

public static void setLambdas(double lambda_sourceGivenTargetLexProb,
                              double lambda_targetGivenSourceLexProb,
                              double lambda_sourceGivenTargetPhraseProb,
                              double lambda_targetGivenSourcePhraseProb,
                              double lambda_languageModel,
                              double lambda_wordPenalty,
                              double lambda_phrasePenalty,
                              double lambda_glue)

getLexLogProbSourceGivenTarget

public double getLexLogProbSourceGivenTarget()

Returns:

the lexLogProbSourceGivenTarget

getLexLogProbTargetGivenSource

public double getLexLogProbTargetGivenSource()

Returns:

the lexLogProbTargetGivenSource

getRuleLogProbSourceGivenTarget

public double getRuleLogProbSourceGivenTarget()

Returns:

the ruleLogProbSourceGivenTarget

getPhrasePenalty

public double getPhrasePenalty()

getRuleLogProbTargetGivenSource

public double getRuleLogProbTargetGivenSource()

Returns:

the ruleLogProbTargetGivenSource

getLanguageModelProb

public double getLanguageModelProb()

Returns:

the languageModelProb

setLanguageModelProb

public void setLanguageModelProb(double languageModelProb)

Set the languageModelProb

Parameters:

languageModelProb -

getNumberOfWords

public int getNumberOfWords()

toString

public String toString()

Overrides:

toString in class Object

getScore

public double getScore()

Return a score.

Returns:

a score

toArray

public double[] toArray()

Return the set of feature values as an array. This is required for minimum error rate training.

Returns:

the set of feature values as an array