#include <wb-word-cluster.h>

Public Member Functions
	WordCluster_t (int nClass, char *pathRes=NULL)

	~WordCluster_t (void)

void	WriteRes (const char *path)

void	ReadRes (const char *path)

void	Reverse (int *pGram)

void	InitCount (const char path, const char path_init_res=NULL)

void	UpdateCount (Array< int > &aCountBuf)

void	CountAdd (Array< int > &aCountBuf, LHash< int, int > &hash, int key, int count)

void	CountAdd (Array< int > &aCountBuf, Trie< int, int > &hash, int *pKey, int nLen, int count)

void	CountAdd (VecShell< int > &aCountBuf, LHash< int, int > &hash, int key, int count)

void	CountAdd (VecShell< int > &aCountBuf, Trie< int, int > &hash, int *pKey, int nLen, int count)

void	CopyCountToThreads (Array< int > &aCountBuf)

void	MoveWord (VecShell< int > vCountBuf, VecShell< int > vMap, int nWord, bool bOut=true)
	move word in/out of a class and update the counts More...

void	ExchangeWord (VecShell< int > vCountBuf, VecShell< int > vMap, int nWord, int nToClass)
	exchange the nWord form m_aClass[nWord] to nToClass More...

void	Cluster (int nMaxTime=-1)
	cluster More...

double	LogLikelihood (VecShell< int > vCountBuf)
	claculate the Loglikelihood More...

void	SimpleCluster ()
	ʹ�ó��Ƶ�ʽ��м򵥵ķ��࣬��Ҫ�� More...

Public Attributes
LHash< int, int >	m_word_count
	N(w) index the word unigram count. More...

Trie< int, int >	m_wgram_count
	N(w,v) word bigram count. More...

Trie< int, int >	m_inv_wgram_count
	N(v,w) inverse word bigram count. More...

LHash< int, int >	m_class
	index the class More...

Trie< int, int >	m_class_gram
	(g(w), g(v)) the index of the class ngram More...

Trie< int, int >	m_word_class_gram
	(w,g) the word-class ngram More...

Trie< int, int >	m_class_word_gram
	(g,w) the class-word ngram More...

Mat< int >	m_tCountBuf
	the count buffer for each threads More...

Mat< int >	m_tMap
	map the word to correspond class at each thread More...

Array< int >	m_mCountBuf
	the count buffer in main threads More...

Array< int >	m_mMap
	the final g(w) More...

int	m_nClassNum
	the maximum class number More...

int	m_nVocabSize
	word number, i.e. the maximum word-id + 1 More...

int	m_nSentNum
	total sentence number More...

int	m_nUnigramNum

int	m_nBigramNum

double	m_dWordLogSum

String	m_pathRes
	the result file, [ w g(w) ] More...

Detailed Description

Definition at line 111 of file wb-word-cluster.h.

Constructor & Destructor Documentation

§ WordCluster_t()

wb::WordCluster_t::WordCluster_t	(	int	nClass,
		char *	pathRes = `NULL`
	)

inline

Definition at line 147 of file wb-word-cluster.h.

§ ~WordCluster_t()

wb::WordCluster_t::~WordCluster_t ( void )

inline

Definition at line 160 of file wb-word-cluster.h.

Member Function Documentation

§ Cluster()

void wb::WordCluster_t::Cluster ( int nMaxTime = -1 )

cluster

Definition at line 865 of file wb-word-cluster.cpp.

§ CopyCountToThreads()

void wb::WordCluster_t::CopyCountToThreads ( Array< int > & aCountBuf )

Definition at line 763 of file wb-word-cluster.cpp.

§ CountAdd() [1/4]

void wb::WordCluster_t::CountAdd	(	Array< int > &	aCountBuf,
		LHash< int, int > &	hash,
		int	key,
		int	count
	)

Definition at line 725 of file wb-word-cluster.cpp.

§ CountAdd() [2/4]

void wb::WordCluster_t::CountAdd	(	Array< int > &	aCountBuf,
		Trie< int, int > &	hash,
		int *	pKey,
		int	nLen,
		int	count
	)

Definition at line 735 of file wb-word-cluster.cpp.

§ CountAdd() [3/4]

void wb::WordCluster_t::CountAdd	(	VecShell< int > &	aCountBuf,
		LHash< int, int > &	hash,
		int	key,
		int	count
	)

Definition at line 745 of file wb-word-cluster.cpp.

§ CountAdd() [4/4]

void wb::WordCluster_t::CountAdd	(	VecShell< int > &	aCountBuf,
		Trie< int, int > &	hash,
		int *	pKey,
		int	nLen,
		int	count
	)

Definition at line 754 of file wb-word-cluster.cpp.

§ ExchangeWord()

void wb::WordCluster_t::ExchangeWord	(	VecShell< int >	vCountBuf,
		VecShell< int >	vMap,
		int	nWord,
		int	nToClass
	)

exchange the nWord form m_aClass[nWord] to nToClass

Definition at line 856 of file wb-word-cluster.cpp.

§ InitCount()

void wb::WordCluster_t::InitCount	(	const char *	path,
		const char *	path_init_res = `NULL`
	)

Definition at line 562 of file wb-word-cluster.cpp.

§ LogLikelihood()

double wb::WordCluster_t::LogLikelihood ( VecShell< int > vCountBuf )

claculate the Loglikelihood

Definition at line 981 of file wb-word-cluster.cpp.

§ MoveWord()

void wb::WordCluster_t::MoveWord	(	VecShell< int >	vCountBuf,
		VecShell< int >	vMap,
		int	nWord,
		bool	bOut = `true`
	)

move word in/out of a class and update the counts

class unigram

class bigram

< the inverse of class-word pairs

Definition at line 778 of file wb-word-cluster.cpp.

§ ReadRes()

void wb::WordCluster_t::ReadRes ( const char * path )

Definition at line 551 of file wb-word-cluster.cpp.

§ Reverse()

void wb::WordCluster_t::Reverse ( int * pGram )

inline

Definition at line 164 of file wb-word-cluster.h.

§ SimpleCluster()

void wb::WordCluster_t::SimpleCluster ( )

ʹ�ó��Ƶ�ʽ��м򵥵ķ��࣬��Ҫ��

< �Դ�Ƶ��ƽ��

Definition at line 1024 of file wb-word-cluster.cpp.

§ UpdateCount()

void wb::WordCluster_t::UpdateCount ( Array< int > & aCountBuf )

Definition at line 662 of file wb-word-cluster.cpp.

§ WriteRes()

void wb::WordCluster_t::WriteRes ( const char * path )

Definition at line 541 of file wb-word-cluster.cpp.

Member Data Documentation

§ m_class

LHash<int, int> wb::WordCluster_t::m_class

index the class

Definition at line 118 of file wb-word-cluster.h.

§ m_class_gram

Trie<int, int> wb::WordCluster_t::m_class_gram

(g(w), g(v)) the index of the class ngram

Definition at line 119 of file wb-word-cluster.h.

§ m_class_word_gram

Trie<int, int> wb::WordCluster_t::m_class_word_gram

(g,w) the class-word ngram

Definition at line 121 of file wb-word-cluster.h.

§ m_dWordLogSum

double wb::WordCluster_t::m_dWordLogSum

Definition at line 136 of file wb-word-cluster.h.

§ m_inv_wgram_count

Trie<int, int> wb::WordCluster_t::m_inv_wgram_count

N(v,w) inverse word bigram count.

Definition at line 116 of file wb-word-cluster.h.

§ m_mCountBuf

Array<int> wb::WordCluster_t::m_mCountBuf

the count buffer in main threads

Definition at line 126 of file wb-word-cluster.h.

§ m_mMap

Array<int> wb::WordCluster_t::m_mMap

the final g(w)

Definition at line 127 of file wb-word-cluster.h.

§ m_nBigramNum

int wb::WordCluster_t::m_nBigramNum

Definition at line 135 of file wb-word-cluster.h.

§ m_nClassNum

int wb::WordCluster_t::m_nClassNum

the maximum class number

Definition at line 130 of file wb-word-cluster.h.

§ m_nSentNum

int wb::WordCluster_t::m_nSentNum

total sentence number

Definition at line 132 of file wb-word-cluster.h.

§ m_nUnigramNum

int wb::WordCluster_t::m_nUnigramNum

Definition at line 134 of file wb-word-cluster.h.

§ m_nVocabSize

int wb::WordCluster_t::m_nVocabSize

word number, i.e. the maximum word-id + 1

Definition at line 131 of file wb-word-cluster.h.

§ m_pathRes

String wb::WordCluster_t::m_pathRes

the result file, [ w g(w) ]

Definition at line 138 of file wb-word-cluster.h.

§ m_tCountBuf

Mat<int> wb::WordCluster_t::m_tCountBuf

the count buffer for each threads

Definition at line 123 of file wb-word-cluster.h.

§ m_tMap

Mat<int> wb::WordCluster_t::m_tMap

map the word to correspond class at each thread

Definition at line 124 of file wb-word-cluster.h.

§ m_wgram_count

Trie<int, int> wb::WordCluster_t::m_wgram_count

N(w,v) word bigram count.

Definition at line 115 of file wb-word-cluster.h.

§ m_word_class_gram

Trie<int, int> wb::WordCluster_t::m_word_class_gram

(w,g) the word-class ngram

Definition at line 120 of file wb-word-cluster.h.

§ m_word_count

LHash<int, int> wb::WordCluster_t::m_word_count

N(w) index the word unigram count.

Definition at line 114 of file wb-word-cluster.h.

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

C:/Users/zhang/spmi/SPMILM/tools/trf/src/base/wb-word-cluster.h
C:/Users/zhang/spmi/SPMILM/tools/trf/src/base/wb-word-cluster.cpp

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

§ WordCluster_t()

§ ~WordCluster_t()

Member Function Documentation

§ Cluster()

§ CopyCountToThreads()

§ CountAdd() [1/4]

§ CountAdd() [2/4]

§ CountAdd() [3/4]

§ CountAdd() [4/4]

§ ExchangeWord()

§ InitCount()

§ LogLikelihood()

§ MoveWord()

§ ReadRes()

§ Reverse()

§ SimpleCluster()

§ UpdateCount()

§ WriteRes()

Member Data Documentation

§ m_class

§ m_class_gram

§ m_class_word_gram

§ m_dWordLogSum

§ m_inv_wgram_count

§ m_mCountBuf

§ m_mMap

§ m_nBigramNum

§ m_nClassNum

§ m_nSentNum

§ m_nUnigramNum

§ m_nVocabSize

§ m_pathRes

§ m_tCountBuf

§ m_tMap

§ m_wgram_count

§ m_word_class_gram

§ m_word_count