A class to represent a candidate in a set of candidates that constitutes a training instance for a reranker. More...

#include <candidate.H>

Classes
class	Comparator
	An inner interface specifying comparison between two Candidate instances. More...

Public Member Functions
	Candidate (int index, double loss, double baseline_score, int num_words, const string &raw_data)
	Constructor for a candidate without features. More...

	Candidate (int index, double loss, double baseline_score, int num_words, const string &raw_data, const FeatureVector< int, double > &features, const FeatureVector< string, double > &symbolic_features)
	Constructor for a candidate with "compiled" features. More...

virtual	~Candidate ()
	Destroys this candidate. More...

int	index () const
	Returns the index of this candidate relative to the other candidates. More...

int	num_errors () const
	Returns the number of word errors of this candidate relative to a reference string. More...

int	num_correct () const
	Returns the number of correct words of this candidate relative to a reference string. More...

double	loss () const
	Returns the loss of this candidate. More...

double	score () const
	Returns the reranker’s score for this candidate. More...

double	baseline_score () const
	Returns the baseline model score for this candidate. More...

int	num_words () const
	Returns the number of words in this candidate. More...

const FeatureVector< int, double > &	features () const
	Returns the feature vector for this candidate. More...

const FeatureVector< string, double > &	symbolic_features () const
	Returns the symbolic feature vector for this candidate. More...

const string &	raw_data () const
	Returns the raw data (typically the sentence) for this candidate. More...

bool	compiled () const
	Returns whether this candidate’s symbolic features have been compiled. More...

void	set_raw_data (const string &raw_data)
	Sets the raw data (typically the sentence) for this candidate). More...

void	set_score (double score)
	Sets the score of this candidate. More...

void	set_baseline_score (double baseline_score)
	Sets the score of this candidate. More...

void	set_loss (double loss)
	Sets the loss of this candidate. More...

bool	Compile (Symbols *symbols, bool clear_features=false, bool clear_symbolic_features=true, bool force=false)
	Compiles any symbolic features in this candidate. More...

void	Decompile (Symbols *symbols, bool clear_symbolic_features=false, bool clear_features=true, bool force=false)
	Decompiles any non-symbolic features in this candidate. More...

string	to_string ()
	Returns a the string that would be output by operator<<. More...

Friends
class	FeatureExtractor

class	AbstractFileBackedFeatureExtractor

ostream &	operator<< (ostream &os, const Candidate &c)
	Outputs a human-readable string version of this Candidate instance to the specified ostream. More...

Detailed Description

A class to represent a candidate in a set of candidates that constitutes a training instance for a reranker.

Definition at line 60 of file candidate.H.

Constructor & Destructor Documentation

reranker::Candidate::Candidate	(	int	index,
		double	loss,
		double	baseline_score,
		int	num_words,
		const string &	raw_data
	)

inline

Constructor for a candidate without features.

Parameters

index	the index of this candidate within its set
loss	the loss of this candidate
baseline_score	the baseline score of this candidate
num_words	the number of words of this candidate
raw_data	the original string for this candidate

Definition at line 71 of file candidate.H.

reranker::Candidate::Candidate	(	int	index,
		double	loss,
		double	baseline_score,
		int	num_words,
		const string &	raw_data,
		const FeatureVector< int, double > &	features,
		const FeatureVector< string, double > &	symbolic_features
	)

inline

Constructor for a candidate with "compiled" features.

Parameters

index	the index of this candidate within its set
loss	the loss of this candidate
baseline_score	the baseline score of this candidate
num_words	the number of words of this candidate
raw_data	the original string for this candidate
features	the features for this candidate
symbolic_features	the symbolic features for this candidate

Definition at line 90 of file candidate.H.

virtual reranker::Candidate::~Candidate ( )

inlinevirtual

Destroys this candidate.

Definition at line 105 of file candidate.H.

Member Function Documentation

double reranker::Candidate::baseline_score ( ) const

inline

Returns the baseline model score for this candidate.

Definition at line 133 of file candidate.H.

bool reranker::Candidate::Compile	(	Symbols *	symbols,
		bool	clear_features = `false`,
		bool	clear_symbolic_features = `true`,
		bool	force = `false`
	)

Compiles any symbolic features in this candidate.

Feature weights for symbolic features will be added to those for any features already specified with int uid’s.

Parameters

symbols	the map from symbols (string instances) to unique integer id’s
clear_features	whether to clear the “normal” feature vector *prior to* compiling symbolic features
clear_symbolic_features	whether to clear the symbolic feature vector *after* updating the “regular” feature vector (to save space)
force	forces feature compilation, even if this method has been previously invoked

Returns: whether any symbolic features were actually compiled by this method invocation

Definition at line 42 of file candidate.C.

bool reranker::Candidate::compiled ( ) const

inline

Returns whether this candidate’s symbolic features have been compiled.

See Also: Compile

Definition at line 147 of file candidate.H.

void reranker::Candidate::Decompile	(	Symbols *	symbols,
		bool	clear_symbolic_features = `false`,
		bool	clear_features = `true`,
		bool	force = `false`
	)

Decompiles any non-symbolic features in this candidate.

Feature weights for non-symbolic features will be added to those for any features already specified with string uid’s.

Parameters

symbols	the map from symbols (string instances) to unique integer id’s
clear_symbolic_features	whether to clear the symbolic feature vector *prior to* decompiling features
clear_features	whether to clear the “normal” feature vector *after* updating the symbolic feature vector (to save space)
force	forces feature decompilation, even if this method has been previously invoked

Definition at line 67 of file candidate.C.

const FeatureVector<int,double>& reranker::Candidate::features ( ) const

inline

Returns the feature vector for this candidate.

Definition at line 137 of file candidate.H.

int reranker::Candidate::index ( ) const

inline

Returns the index of this candidate relative to the other candidates.

Definition at line 121 of file candidate.H.

double reranker::Candidate::loss ( ) const

inline

Returns the loss of this candidate.

Definition at line 129 of file candidate.H.

int reranker::Candidate::num_correct ( ) const

inline

Returns the number of correct words of this candidate relative to a reference string.

Definition at line 127 of file candidate.H.

int reranker::Candidate::num_errors ( ) const

inline

Returns the number of word errors of this candidate relative to a reference string.

Definition at line 124 of file candidate.H.

int reranker::Candidate::num_words ( ) const

inline

Returns the number of words in this candidate.

Definition at line 135 of file candidate.H.

const string& reranker::Candidate::raw_data ( ) const

inline

Returns the raw data (typically the sentence) for this candidate.

Definition at line 143 of file candidate.H.

double reranker::Candidate::score ( ) const

inline

Returns the reranker’s score for this candidate.

Definition at line 131 of file candidate.H.

void reranker::Candidate::set_baseline_score ( double baseline_score )

inline

Sets the score of this candidate.

Parameters

baseline_score the new baseline_score of this candidate

Definition at line 161 of file candidate.H.

void reranker::Candidate::set_loss ( double loss )

inline

Sets the loss of this candidate.

Parameters

loss	the new loss of this candidate.

Definition at line 168 of file candidate.H.

void reranker::Candidate::set_raw_data ( const string & raw_data )

inline

Sets the raw data (typically the sentence) for this candidate).

Definition at line 151 of file candidate.H.

void reranker::Candidate::set_score ( double score )

inline

Sets the score of this candidate.

Parameters

score the new score of this candidate

Definition at line 156 of file candidate.H.

const FeatureVector<string,double>& reranker::Candidate::symbolic_features ( ) const

inline

Returns the symbolic feature vector for this candidate.

Definition at line 139 of file candidate.H.

string reranker::Candidate::to_string ( )

inline

Returns a the string that would be output by operator<<.

Definition at line 232 of file candidate.H.

Friends And Related Function Documentation

friend class AbstractFileBackedFeatureExtractor

friend

Definition at line 63 of file candidate.H.

friend class FeatureExtractor

friend

Definition at line 62 of file candidate.H.

ostream& operator<<	(	ostream &	os,
		const Candidate &	c
	)

friend

Outputs a human-readable string version of this Candidate instance to the specified ostream.

Parameters

os	the output stream to which to output a human-readable version of this candidate
c	the candidate to be output to the specified output stream

Definition at line 220 of file candidate.H.

The documentation for this class was generated from the following files:

Classes

Public Member Functions

Friends

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Friends And Related Function Documentation