moot::mootHMMTrainer Class Reference

High-level class to gather training data for a mootHMM or mootCHMM.

Collaboration diagram for moot::mootHMMTrainer:

Public Types
Training types
typedef mootNgrams::Ngram	Ngram
typedef mootNgrams::NgramCount	CountT
typedef set< mootTagString >	TagSet

Public Member Functions
Constructor / destructor
	mootHMMTrainer (void)
	~mootHMMTrainer (void)
Reset / Clear
void	clear (void)
Top-level training methods
bool	train_from_reader (TokenReader *reader)
bool	train_from_stream (FILE *in=stdin, const string &srcname="(unknown)")
bool	train_from_file (const string &filename)
bool	train_finish (void)
Mid-level training methods
void	train_init (void)
void	train_bos (void)
void	train_token (const mootToken &curtok)
void	train_eos (void)
Warnings / Errors
void	carp (const char *fmt,...)

Public Attributes
Training data
mootNgrams	ngrams
mootLexfreqs	lexfreqs
mootClassfreqs	lcfreqs
mootTaster	taster
Flags
bool	want_ngrams
bool	want_lexfreqs
bool	want_classfreqs
bool	want_flavors
Pragmatic constants
mootTagString	eos_tag

Protected Attributes
Runtime training state
Ngram	ng
bool	last_was_eos

Member Typedef Documentation

Ngram

typedef mootNgrams::Ngram moot::mootHMMTrainer::Ngram

Type for an N-gram

CountT

typedef mootNgrams::NgramCount moot::mootHMMTrainer::CountT

Type for counts

TagSet

typedef set<mootTagString> moot::mootHMMTrainer::TagSet

Type for current tag-sets

Constructor & Destructor Documentation

mootHMMTrainer()

moot::mootHMMTrainer::mootHMMTrainer ( void  )

inline

Default constructor

~mootHMMTrainer()

moot::mootHMMTrainer::~mootHMMTrainer ( void  )

inline

Default destructor

Member Function Documentation

clear()

void moot::mootHMMTrainer::clear ( void  )

inline

Reset / clear the object.

References moot::mootLexfreqs::clear(), moot::mootClassfreqs::clear(), moot::mootNgrams::clear(), and moot::mootTaster::set_default_rules().

train_from_reader()

bool moot::mootHMMTrainer::train_from_reader

(

TokenReader *

reader

)

Gather training data using TokenIO layer

train_from_stream()

bool moot::mootHMMTrainer::train_from_stream	(	FILE *	in = stdin,
		const string &	srcname = "(unknown)"
	)

Gather training data from a native text-format C-stream

train_from_file()

bool moot::mootHMMTrainer::train_from_file

(

const string &

filename

)

Gather training data from a file using mootTaggerLexer

train_finish()

bool moot::mootHMMTrainer::train_finish

(

void

)

Finish training and compute "special" pseudo-frequencies (e.g. , flavors, etc.)

train_init()

void moot::mootHMMTrainer::train_init

(

void

)

Initialize training data

train_bos()

void moot::mootHMMTrainer::train_bos

(

void

)

Initialize data for training a new sentence

train_token()

void moot::mootHMMTrainer::train_token

(

const mootToken &

curtok

)

Gather training information for a single token, using mootToken interface

train_eos()

void moot::mootHMMTrainer::train_eos

(

void

)

Gather training information for a sentence boundary.

carp()

void moot::mootHMMTrainer::carp	(	const char *	fmt,
			...
	)

Error reporting

Member Data Documentation

ngrams

mootNgrams moot::mootHMMTrainer::ngrams

Raw n-gram frequency data

lexfreqs

mootLexfreqs moot::mootHMMTrainer::lexfreqs

Raw lexical frequency data

lcfreqs

mootClassfreqs moot::mootHMMTrainer::lcfreqs

Raw lexical-class frequency data

taster

mootTaster moot::mootHMMTrainer::taster

Heuristic token classifier (default: built-in rules)

want_ngrams

bool moot::mootHMMTrainer::want_ngrams

Whether to gather n-gram frequency data

want_lexfreqs

bool moot::mootHMMTrainer::want_lexfreqs

Whether to gather lexical frequency data

want_classfreqs

bool moot::mootHMMTrainer::want_classfreqs

Whether to gather lexical-class frequency data

want_flavors

bool moot::mootHMMTrainer::want_flavors

Whether to gather lexical-flavor information

eos_tag

mootTagString moot::mootHMMTrainer::eos_tag

String indicating end-of-sentence: this is usually __$

ng

Ngram moot::mootHMMTrainer::ng

protected

Current n-gram window

last_was_eos

bool moot::mootHMMTrainer::last_was_eos

protected

Stupid hack

---

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

• mootHMMTrainer.h