mosesdecoder/moses/DummyScoreProducers.h

114 lines
2.9 KiB
C
Raw Normal View History

// $Id$
#ifndef moses_DummyScoreProducers_h
#define moses_DummyScoreProducers_h
Feature function overhaul. Each feature function is computed in one of three ways: 1) Stateless feature functions from the phrase table/generation table: these are computed when the TranslationOption is created. They become part of the ScoreBreakdown object contained in the TranslationOption and are added to the feature value vector when a hypothesis is extended. 2) Stateless feature functions that are computed during state exploration. Currently, only WordPenalty falls into this category, but these functions implement a method Evaluate which do does not receive a Hypothesis or any contextual information. 3) Stateful feature functions: these features receive the arc information (translation option), compute some value and then return some context information. The context information created by a particular feature function is passed back to it as the previous context when a hypothesis originating at the node where the previous edge terminates is created. States in the search space may be recombined if the context information is identical. The context information must be stored in an object implementing the FFState interface. TODO: 1) the command line interface / MERT interface needs to go to named parameters that are otherwise opaque 2) StatefulFeatureFunction's Evaluate method should just take a TranslationOption and a context object. It is not good that it takes a hypothesis, because then people may be tempted to access information about the "previous" hypothesis without "declaring" this dependency. 3) Future cost estimates should be handled using feature functions. All stateful feature functions need some kind of future cost estimate. 4) Philipp's poor-man's cube pruning is broken. git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@2087 1f5c12ca-751b-0410-a591-d2e778427230
2009-02-06 18:43:06 +03:00
#include "FeatureFunction.h"
namespace Moses
{
class WordsRange;
/** Calculates Distortion scores
*/
class DistortionScoreProducer : public StatefulFeatureFunction
{
public:
DistortionScoreProducer() : StatefulFeatureFunction("Distortion", 1) {}
float CalculateDistortionScore(const Hypothesis& hypo,
const WordsRange &prev, const WordsRange &curr, const int FirstGapPosition) const;
std::string GetScoreProducerWeightShortName(unsigned) const;
virtual const FFState* EmptyHypothesisState(const InputType &input) const;
Feature function overhaul. Each feature function is computed in one of three ways: 1) Stateless feature functions from the phrase table/generation table: these are computed when the TranslationOption is created. They become part of the ScoreBreakdown object contained in the TranslationOption and are added to the feature value vector when a hypothesis is extended. 2) Stateless feature functions that are computed during state exploration. Currently, only WordPenalty falls into this category, but these functions implement a method Evaluate which do does not receive a Hypothesis or any contextual information. 3) Stateful feature functions: these features receive the arc information (translation option), compute some value and then return some context information. The context information created by a particular feature function is passed back to it as the previous context when a hypothesis originating at the node where the previous edge terminates is created. States in the search space may be recombined if the context information is identical. The context information must be stored in an object implementing the FFState interface. TODO: 1) the command line interface / MERT interface needs to go to named parameters that are otherwise opaque 2) StatefulFeatureFunction's Evaluate method should just take a TranslationOption and a context object. It is not good that it takes a hypothesis, because then people may be tempted to access information about the "previous" hypothesis without "declaring" this dependency. 3) Future cost estimates should be handled using feature functions. All stateful feature functions need some kind of future cost estimate. 4) Philipp's poor-man's cube pruning is broken. git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@2087 1f5c12ca-751b-0410-a591-d2e778427230
2009-02-06 18:43:06 +03:00
virtual FFState* Evaluate(
const Hypothesis& cur_hypo,
const FFState* prev_state,
ScoreComponentCollection* accumulator) const;
virtual FFState* EvaluateChart(
const ChartHypothesis& /* cur_hypo */,
int /* featureID - used to index the state in the previous hypotheses */,
ScoreComponentCollection*) const {
CHECK(0); // feature function not valid in chart decoder
return NULL;
}
};
/** Doesn't do anything but provide a key into the global
* score array to store the word penalty in.
*/
class WordPenaltyProducer : public StatelessFeatureFunction
{
public:
WordPenaltyProducer() : StatelessFeatureFunction("WordPenalty",1) {}
std::string GetScoreProducerWeightShortName(unsigned) const;
Feature function overhaul. Each feature function is computed in one of three ways: 1) Stateless feature functions from the phrase table/generation table: these are computed when the TranslationOption is created. They become part of the ScoreBreakdown object contained in the TranslationOption and are added to the feature value vector when a hypothesis is extended. 2) Stateless feature functions that are computed during state exploration. Currently, only WordPenalty falls into this category, but these functions implement a method Evaluate which do does not receive a Hypothesis or any contextual information. 3) Stateful feature functions: these features receive the arc information (translation option), compute some value and then return some context information. The context information created by a particular feature function is passed back to it as the previous context when a hypothesis originating at the node where the previous edge terminates is created. States in the search space may be recombined if the context information is identical. The context information must be stored in an object implementing the FFState interface. TODO: 1) the command line interface / MERT interface needs to go to named parameters that are otherwise opaque 2) StatefulFeatureFunction's Evaluate method should just take a TranslationOption and a context object. It is not good that it takes a hypothesis, because then people may be tempted to access information about the "previous" hypothesis without "declaring" this dependency. 3) Future cost estimates should be handled using feature functions. All stateful feature functions need some kind of future cost estimate. 4) Philipp's poor-man's cube pruning is broken. git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@2087 1f5c12ca-751b-0410-a591-d2e778427230
2009-02-06 18:43:06 +03:00
virtual void Evaluate(
const PhraseBasedFeatureContext& context,
ScoreComponentCollection* accumulator) const;
virtual void EvaluateChart(
2012-09-21 14:56:01 +04:00
const ChartBasedFeatureContext& context,
ScoreComponentCollection* accumulator) const
{
//required but does nothing.
}
};
/** unknown word penalty */
class UnknownWordPenaltyProducer : public StatelessFeatureFunction
{
public:
UnknownWordPenaltyProducer() : StatelessFeatureFunction("!UnknownWordPenalty",1) {}
std::string GetScoreProducerWeightShortName(unsigned) const;
Feature function overhaul. Each feature function is computed in one of three ways: 1) Stateless feature functions from the phrase table/generation table: these are computed when the TranslationOption is created. They become part of the ScoreBreakdown object contained in the TranslationOption and are added to the feature value vector when a hypothesis is extended. 2) Stateless feature functions that are computed during state exploration. Currently, only WordPenalty falls into this category, but these functions implement a method Evaluate which do does not receive a Hypothesis or any contextual information. 3) Stateful feature functions: these features receive the arc information (translation option), compute some value and then return some context information. The context information created by a particular feature function is passed back to it as the previous context when a hypothesis originating at the node where the previous edge terminates is created. States in the search space may be recombined if the context information is identical. The context information must be stored in an object implementing the FFState interface. TODO: 1) the command line interface / MERT interface needs to go to named parameters that are otherwise opaque 2) StatefulFeatureFunction's Evaluate method should just take a TranslationOption and a context object. It is not good that it takes a hypothesis, because then people may be tempted to access information about the "previous" hypothesis without "declaring" this dependency. 3) Future cost estimates should be handled using feature functions. All stateful feature functions need some kind of future cost estimate. 4) Philipp's poor-man's cube pruning is broken. git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@2087 1f5c12ca-751b-0410-a591-d2e778427230
2009-02-06 18:43:06 +03:00
virtual bool ComputeValueInTranslationOption() const;
void Evaluate( const PhraseBasedFeatureContext& context,
ScoreComponentCollection* accumulator) const
{
//do nothing - not a real feature
}
void EvaluateChart(
2012-09-21 14:56:01 +04:00
const ChartBasedFeatureContext& context,
ScoreComponentCollection* accumulator) const
{
//do nothing - not a real feature
}
Feature function overhaul. Each feature function is computed in one of three ways: 1) Stateless feature functions from the phrase table/generation table: these are computed when the TranslationOption is created. They become part of the ScoreBreakdown object contained in the TranslationOption and are added to the feature value vector when a hypothesis is extended. 2) Stateless feature functions that are computed during state exploration. Currently, only WordPenalty falls into this category, but these functions implement a method Evaluate which do does not receive a Hypothesis or any contextual information. 3) Stateful feature functions: these features receive the arc information (translation option), compute some value and then return some context information. The context information created by a particular feature function is passed back to it as the previous context when a hypothesis originating at the node where the previous edge terminates is created. States in the search space may be recombined if the context information is identical. The context information must be stored in an object implementing the FFState interface. TODO: 1) the command line interface / MERT interface needs to go to named parameters that are otherwise opaque 2) StatefulFeatureFunction's Evaluate method should just take a TranslationOption and a context object. It is not good that it takes a hypothesis, because then people may be tempted to access information about the "previous" hypothesis without "declaring" this dependency. 3) Future cost estimates should be handled using feature functions. All stateful feature functions need some kind of future cost estimate. 4) Philipp's poor-man's cube pruning is broken. git-svn-id: https://mosesdecoder.svn.sourceforge.net/svnroot/mosesdecoder/trunk@2087 1f5c12ca-751b-0410-a591-d2e778427230
2009-02-06 18:43:06 +03:00
bool ComputeValueInTranslationTable() const {return true;}
};
class MetaFeatureProducer : public StatelessFeatureFunction
{
public:
MetaFeatureProducer(std::string shortName) : StatelessFeatureFunction("MetaFeature_"+shortName,1), m_shortName(shortName) {}
std::string m_shortName;
std::string GetScoreProducerWeightShortName(unsigned) const;
void Evaluate(const PhraseBasedFeatureContext& context,
ScoreComponentCollection* accumulator) const {
//do nothing - not a real feature
}
void EvaluateChart(const ChartBasedFeatureContext& context,
ScoreComponentCollection*) const {
//do nothing - not a real feature
}
};
}
#endif