A shift-reduce constituency parser. More...

#include <sr_parser.h>

Classes
class	exception
	Exception thrown during parser actions. More...

class	state_analyzer
	Analyzer responsible for converting a parser state to a feature_vector. More...

struct	training_batch
	A training batch. More...

class	training_data
	Training data for the parser. More...

struct	training_options
	Training options required for learning a parser model. More...

Public Types
enum	training_algorithm { EARLY_TERMINATION, BEAM_SEARCH }
	The set of training algorithms available for the parser.

using	feature_vector = std::unordered_map< std::string, float >
	Sparse vector representation of a state's features.

using	weight_vector = util::sparse_vector< trans_id, float >
	A single weight vector for a specific transition.

using	weight_vectors = std::unordered_map< std::string, weight_vector >
	A collection of weight vectors by feature type.

Public Member Functions
	sr_parser ()=default
	Default constructor.

	sr_parser (const std::string &prefix)
	Loads a pre-trained parser from a prefix. More...

parse_tree	parse (const sequence::sequence &sentence) const
	Parses a POS-tagged sentence (represented as a sequence::sequence). More...

void	train (std::vector< parse_tree > &trees, training_options options)
	Trains a model on the given parse trees using the supplied training options. More...

void	save (const std::string &prefix) const

Private Types
using	scored_trans = std::pair< trans_id, float >

Private Member Functions
void	load (const std::string &prefix)

std::tuple< weight_vectors, uint64_t, uint64_t >	train_batch (training_batch batch, parallel::thread_pool &pool, const training_options &options)
	Calculates a weight update on a given batch of training trees. More...

std::pair< uint64_t, uint64_t >	train_instance (const parse_tree &tree, const std::vector< trans_id > &transitions, const training_options &options, weight_vectors &update) const
	Calculates a weight update on a single tree. More...

std::pair< uint64_t, uint64_t >	train_early_termination (const parse_tree &tree, const std::vector< trans_id > &transitions, weight_vectors &update) const
	Calculates a weight update on a single tree, using the greedy early termination training strategy. More...

std::pair< uint64_t, uint64_t >	train_beam_search (const parse_tree &tree, const std::vector< trans_id > &transitions, const training_options &options, weight_vectors &update) const
	Calculates a weight update on a single tree, using beam search. More...

trans_id	best_transition (const feature_vector &features, const state &state, bool check_legality=false) const
	Computes the most likely transition according to the current model. More...

std::vector< scored_trans >	best_transitions (const feature_vector &features, const state &state, size_t num, bool check_legality=false) const
	Computes the \(k\) most likely transitions according to the current model. More...

Private Attributes
transition_map	trans_
	Storage for the ids for each transition.

classify::linear_model< std::string, float, trans_id >	model_
	Storage for the weights for each possible transition.

uint64_t	beam_size_ = 1
	Beam size used during training.

Detailed Description

A shift-reduce constituency parser.

The model is a simple linear classifier learned via the generalized averaged perceptron algorithm that seeks to classify a parser action given a parser state.

See also: http://people.sutd.edu.sg/~yue_zhang/pub/acl13.muhua.pdf; http://www.aclweb.org/anthology/W09-3825

Constructor & Destructor Documentation

meta::parser::sr_parser::sr_parser ( const std::string & prefix )

Loads a pre-trained parser from a prefix.

Parameters

prefix The prefix to load the parser model from

Member Function Documentation

parse_tree meta::parser::sr_parser::parse ( const sequence::sequence & sentence ) const

Parses a POS-tagged sentence (represented as a sequence::sequence).

Parameters

sentence The sentence to be tagged

Returns: the parse tree corresponding to the input sentence

void meta::parser::sr_parser::train	(	std::vector< parse_tree > &	trees,
		training_options	options
	)

Trains a model on the given parse trees using the supplied training options.

Parameters

trees	The full parse trees for training
options	The options used for training

void meta::parser::sr_parser::save ( const std::string & prefix ) const

Parameters

prefix The prefix to store the model in

void meta::parser::sr_parser::load ( const std::string & prefix )

private

Parameters

prefix The prefix to load the model from

auto meta::parser::sr_parser::train_batch	(	training_batch	batch,
		parallel::thread_pool &	pool,
		const training_options &	options
	)

private

Calculates a weight update on a given batch of training trees.

Parameters

batch	The batch to learn on
pool	The thread pool to use for parsing the batch in parallel
options	The training options

Returns: a 3-tuple (update, correct actions, incorrect actions)

std::pair< uint64_t, uint64_t > meta::parser::sr_parser::train_instance	(	const parse_tree &	tree,
		const std::vector< trans_id > &	transitions,
		const training_options &	options,
		weight_vectors &	update
	)		const

private

Calculates a weight update on a single tree.

Parameters

tree	The training tree
transitions	The correct transitions for parsing this tree
options	The training options
update	The weight vector to place the update in

Returns: (correct actions, incorrect actions)

std::pair< uint64_t, uint64_t > meta::parser::sr_parser::train_early_termination	(	const parse_tree &	tree,
		const std::vector< trans_id > &	transitions,
		weight_vectors &	update
	)		const

private

Calculates a weight update on a single tree, using the greedy early termination training strategy.

Parameters

tree	The training tree
transitions	The correct transitions for parsing this tree
options	The training options
update	The weight vector to place the update in

Returns: (correct actions, incorrect actions)

std::pair< uint64_t, uint64_t > meta::parser::sr_parser::train_beam_search	(	const parse_tree &	tree,
		const std::vector< trans_id > &	transitions,
		const training_options &	options,
		weight_vectors &	update
	)		const

private

Calculates a weight update on a single tree, using beam search.

Parameters

tree	The training tree
transitions	The correct transitions for parsing this tree
options	The training options
update	The weight vector to place the update in

Returns: (correct actions, incorrect actions)

auto meta::parser::sr_parser::best_transition	(	const feature_vector &	features,
		const state &	state,
		bool	check_legality = `false`
	)		const

private

Computes the most likely transition according to the current model.

Parameters

features	The feature vector representation for the current state
state	The current state
check_legality	Whether or not to limit the transitions to only those that are "legal" according to the constraints given for each transition

auto meta::parser::sr_parser::best_transitions	(	const feature_vector &	features,
		const state &	state,
		size_t	num,
		bool	check_legality = `false`
	)		const

private

Computes the \(k\) most likely transitions according to the current model.

Parameters

features	The feature vector representation for the current state
state	The current state
check_legality	Whether or not to limit the transitions to only those that are "legal" according to the constraints given for each transition

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

/home/chase/projects/meta/include/parser/sr_parser.h
/home/chase/projects/meta/src/parser/sr_parser.cpp

Classes

Public Types

Public Member Functions

Private Types

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation