2014-12-21 09:59:07 +00:00
|
|
|
from os import path
|
2014-12-23 00:40:32 +00:00
|
|
|
import json
|
2014-12-31 08:40:59 +00:00
|
|
|
import os
|
|
|
|
import shutil
|
2014-12-23 00:40:32 +00:00
|
|
|
|
2014-12-30 10:20:15 +00:00
|
|
|
from libc.string cimport memset
|
|
|
|
|
|
|
|
from cymem.cymem cimport Address
|
2014-12-31 08:40:59 +00:00
|
|
|
from thinc.typedefs cimport atom_t, weight_t
|
2014-12-21 09:59:07 +00:00
|
|
|
|
2014-12-23 02:18:59 +00:00
|
|
|
from ..typedefs cimport univ_tag_t
|
2014-12-21 09:59:07 +00:00
|
|
|
from ..typedefs cimport NO_TAG, ADJ, ADV, ADP, CONJ, DET, NOUN, NUM, PRON, PRT, VERB
|
|
|
|
from ..typedefs cimport X, PUNCT, EOL
|
2014-12-23 02:18:59 +00:00
|
|
|
from ..typedefs cimport id_t
|
2015-01-11 23:26:22 +00:00
|
|
|
from ..structs cimport TokenC, Morphology, LexemeC
|
2014-12-21 09:59:07 +00:00
|
|
|
from ..tokens cimport Tokens
|
2014-12-23 02:18:59 +00:00
|
|
|
from ..morphology cimport set_morph_from_dict
|
2014-12-31 08:40:59 +00:00
|
|
|
from .._ml cimport arg_max
|
|
|
|
|
2014-12-23 00:40:32 +00:00
|
|
|
from .lemmatizer import Lemmatizer
|
2014-12-21 09:59:07 +00:00
|
|
|
|
|
|
|
|
|
|
|
cpdef enum en_person_t:
|
|
|
|
NO_PERSON
|
|
|
|
FIRST
|
|
|
|
SECOND
|
|
|
|
THIRD
|
|
|
|
NON_THIRD
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum en_number_t:
|
|
|
|
NO_NUMBER
|
|
|
|
SINGULAR
|
|
|
|
PLURAL
|
|
|
|
MASS
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum en_gender_t:
|
|
|
|
NO_GENDER
|
|
|
|
MASCULINE
|
|
|
|
FEMININE
|
|
|
|
NEUTER
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum en_case_t:
|
|
|
|
NO_CASE
|
|
|
|
NOMINATIVE
|
|
|
|
GENITIVE
|
|
|
|
ACCUSATIVE
|
|
|
|
REFLEXIVE
|
|
|
|
DEMONYM
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum en_tenspect_t:
|
|
|
|
NO_TENSE
|
|
|
|
BASE_VERB
|
|
|
|
PRESENT
|
|
|
|
PAST
|
|
|
|
PASSIVE
|
|
|
|
ING
|
|
|
|
MODAL
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum misc_t:
|
|
|
|
NO_MISC
|
|
|
|
COMPARATIVE
|
|
|
|
SUPERLATIVE
|
|
|
|
RELATIVE
|
|
|
|
NAME
|
|
|
|
|
|
|
|
|
|
|
|
cpdef enum:
|
2015-01-22 15:08:25 +00:00
|
|
|
P2_orth
|
2014-12-21 09:59:07 +00:00
|
|
|
P2_cluster
|
|
|
|
P2_shape
|
|
|
|
P2_prefix
|
|
|
|
P2_suffix
|
|
|
|
P2_pos
|
|
|
|
P2_lemma
|
|
|
|
|
2015-01-22 15:08:25 +00:00
|
|
|
P1_orth
|
2014-12-21 09:59:07 +00:00
|
|
|
P1_cluster
|
|
|
|
P1_shape
|
|
|
|
P1_prefix
|
|
|
|
P1_suffix
|
|
|
|
P1_pos
|
|
|
|
P1_lemma
|
|
|
|
|
2015-01-22 15:08:25 +00:00
|
|
|
W_orth
|
2014-12-21 09:59:07 +00:00
|
|
|
W_cluster
|
|
|
|
W_shape
|
|
|
|
W_prefix
|
|
|
|
W_suffix
|
|
|
|
W_pos
|
|
|
|
W_lemma
|
|
|
|
|
2015-01-22 15:08:25 +00:00
|
|
|
N1_orth
|
2014-12-21 09:59:07 +00:00
|
|
|
N1_cluster
|
|
|
|
N1_shape
|
|
|
|
N1_prefix
|
|
|
|
N1_suffix
|
|
|
|
N1_pos
|
|
|
|
N1_lemma
|
|
|
|
|
2015-01-22 15:08:25 +00:00
|
|
|
N2_orth
|
2014-12-21 09:59:07 +00:00
|
|
|
N2_cluster
|
|
|
|
N2_shape
|
|
|
|
N2_prefix
|
|
|
|
N2_suffix
|
|
|
|
N2_pos
|
|
|
|
N2_lemma
|
|
|
|
|
|
|
|
N_CONTEXT_FIELDS
|
|
|
|
|
|
|
|
|
|
|
|
POS_TAGS = {
|
|
|
|
'NULL': (NO_TAG, {}),
|
|
|
|
'EOL': (EOL, {}),
|
|
|
|
'CC': (CONJ, {}),
|
|
|
|
'CD': (NUM, {}),
|
|
|
|
'DT': (DET, {}),
|
|
|
|
'EX': (DET, {}),
|
|
|
|
'FW': (X, {}),
|
|
|
|
'IN': (ADP, {}),
|
|
|
|
'JJ': (ADJ, {}),
|
|
|
|
'JJR': (ADJ, {'misc': COMPARATIVE}),
|
|
|
|
'JJS': (ADJ, {'misc': SUPERLATIVE}),
|
|
|
|
'LS': (X, {}),
|
|
|
|
'MD': (VERB, {'tenspect': MODAL}),
|
|
|
|
'NN': (NOUN, {}),
|
|
|
|
'NNS': (NOUN, {'number': PLURAL}),
|
|
|
|
'NNP': (NOUN, {'misc': NAME}),
|
|
|
|
'NNPS': (NOUN, {'misc': NAME, 'number': PLURAL}),
|
|
|
|
'PDT': (DET, {}),
|
|
|
|
'POS': (PRT, {'case': GENITIVE}),
|
|
|
|
'PRP': (NOUN, {}),
|
|
|
|
'PRP$': (NOUN, {'case': GENITIVE}),
|
|
|
|
'RB': (ADV, {}),
|
|
|
|
'RBR': (ADV, {'misc': COMPARATIVE}),
|
|
|
|
'RBS': (ADV, {'misc': SUPERLATIVE}),
|
|
|
|
'RP': (PRT, {}),
|
|
|
|
'SYM': (X, {}),
|
|
|
|
'TO': (PRT, {}),
|
|
|
|
'UH': (X, {}),
|
|
|
|
'VB': (VERB, {}),
|
|
|
|
'VBD': (VERB, {'tenspect': PAST}),
|
|
|
|
'VBG': (VERB, {'tenspect': ING}),
|
|
|
|
'VBN': (VERB, {'tenspect': PASSIVE}),
|
|
|
|
'VBP': (VERB, {'tenspect': PRESENT}),
|
|
|
|
'VBZ': (VERB, {'tenspect': PRESENT, 'person': THIRD}),
|
|
|
|
'WDT': (DET, {'misc': RELATIVE}),
|
|
|
|
'WP': (PRON, {'misc': RELATIVE}),
|
|
|
|
'WP$': (PRON, {'misc': RELATIVE, 'case': GENITIVE}),
|
|
|
|
'WRB': (ADV, {'misc': RELATIVE}),
|
|
|
|
'!': (PUNCT, {}),
|
|
|
|
'#': (PUNCT, {}),
|
|
|
|
'$': (PUNCT, {}),
|
|
|
|
"''": (PUNCT, {}),
|
|
|
|
"(": (PUNCT, {}),
|
|
|
|
")": (PUNCT, {}),
|
|
|
|
"-LRB-": (PUNCT, {}),
|
|
|
|
"-RRB-": (PUNCT, {}),
|
|
|
|
".": (PUNCT, {}),
|
|
|
|
",": (PUNCT, {}),
|
|
|
|
"``": (PUNCT, {}),
|
|
|
|
":": (PUNCT, {}),
|
|
|
|
"?": (PUNCT, {}),
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
POS_TEMPLATES = (
|
2015-01-22 15:08:25 +00:00
|
|
|
(W_orth,),
|
2014-12-21 09:59:07 +00:00
|
|
|
(P1_lemma, P1_pos),
|
|
|
|
(P2_lemma, P2_pos),
|
2015-01-22 15:08:25 +00:00
|
|
|
(N1_orth,),
|
|
|
|
(N2_orth,),
|
2014-12-21 09:59:07 +00:00
|
|
|
|
|
|
|
(W_suffix,),
|
|
|
|
(W_prefix,),
|
|
|
|
|
|
|
|
(P1_pos,),
|
|
|
|
(P2_pos,),
|
|
|
|
(P1_pos, P2_pos),
|
2015-01-22 15:08:25 +00:00
|
|
|
(P1_pos, W_orth),
|
2014-12-21 09:59:07 +00:00
|
|
|
(P1_suffix,),
|
|
|
|
(N1_suffix,),
|
|
|
|
|
|
|
|
(W_shape,),
|
|
|
|
(W_cluster,),
|
|
|
|
(N1_cluster,),
|
|
|
|
(N2_cluster,),
|
|
|
|
(P1_cluster,),
|
|
|
|
(P2_cluster,),
|
|
|
|
)
|
|
|
|
|
|
|
|
|
2014-12-23 02:18:59 +00:00
|
|
|
cdef struct _CachedMorph:
|
|
|
|
Morphology morph
|
|
|
|
int lemma
|
|
|
|
|
|
|
|
|
2014-12-31 08:40:59 +00:00
|
|
|
def setup_model_dir(tag_names, tag_map, templates, model_dir):
|
|
|
|
if path.exists(model_dir):
|
|
|
|
shutil.rmtree(model_dir)
|
|
|
|
os.mkdir(model_dir)
|
|
|
|
config = {
|
|
|
|
'templates': templates,
|
|
|
|
'tag_names': tag_names,
|
|
|
|
'tag_map': tag_map
|
|
|
|
}
|
|
|
|
with open(path.join(model_dir, 'config.json'), 'w') as file_:
|
|
|
|
json.dump(config, file_)
|
|
|
|
|
|
|
|
|
2014-12-30 10:20:15 +00:00
|
|
|
cdef class EnPosTagger:
|
2014-12-27 07:45:16 +00:00
|
|
|
"""A part-of-speech tagger for English"""
|
2014-12-23 00:40:32 +00:00
|
|
|
def __init__(self, StringStore strings, data_dir):
|
2014-12-30 10:20:15 +00:00
|
|
|
self.mem = Pool()
|
2014-12-21 09:59:07 +00:00
|
|
|
model_dir = path.join(data_dir, 'pos')
|
2014-12-23 00:40:32 +00:00
|
|
|
self.strings = strings
|
|
|
|
cfg = json.load(open(path.join(data_dir, 'pos', 'config.json')))
|
2014-12-23 02:18:59 +00:00
|
|
|
self.tag_names = sorted(cfg['tag_names'])
|
2014-12-30 10:20:15 +00:00
|
|
|
self.n_tags = len(self.tag_names)
|
2014-12-23 02:18:59 +00:00
|
|
|
self.tag_map = cfg['tag_map']
|
|
|
|
cdef int n_tags = len(self.tag_names) + 1
|
2014-12-30 10:20:15 +00:00
|
|
|
|
2014-12-31 08:40:59 +00:00
|
|
|
self.model = Model(n_tags, cfg['templates'], model_dir)
|
2014-12-23 02:18:59 +00:00
|
|
|
self._morph_cache = PreshMapArray(n_tags)
|
|
|
|
self.tags = <PosTag*>self.mem.alloc(n_tags, sizeof(PosTag))
|
|
|
|
for i, tag in enumerate(sorted(self.tag_names)):
|
|
|
|
pos, props = self.tag_map[tag]
|
|
|
|
self.tags[i].id = i
|
|
|
|
self.tags[i].pos = pos
|
|
|
|
set_morph_from_dict(&self.tags[i].morph, props)
|
2015-01-03 12:13:00 +00:00
|
|
|
if path.exists(path.join(data_dir, 'tokenizer', 'morphs.json')):
|
|
|
|
self.load_morph_exceptions(json.load(open(path.join(data_dir, 'tokenizer',
|
|
|
|
'morphs.json'))))
|
2014-12-23 02:18:59 +00:00
|
|
|
self.lemmatizer = Lemmatizer(path.join(data_dir, 'wordnet'), NOUN, VERB, ADJ)
|
2014-12-21 09:59:07 +00:00
|
|
|
|
|
|
|
def __call__(self, Tokens tokens):
|
2014-12-27 07:45:16 +00:00
|
|
|
"""Apply the tagger, setting the POS tags onto the Tokens object.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
tokens (Tokens): The tokens to be tagged.
|
|
|
|
"""
|
2014-12-21 09:59:07 +00:00
|
|
|
cdef int i
|
|
|
|
cdef atom_t[N_CONTEXT_FIELDS] context
|
2014-12-31 08:40:59 +00:00
|
|
|
cdef const weight_t* scores
|
2014-12-21 09:59:07 +00:00
|
|
|
for i in range(tokens.length):
|
2015-01-17 05:21:17 +00:00
|
|
|
if tokens.data[i].pos == 0:
|
2014-12-31 08:40:59 +00:00
|
|
|
fill_context(context, i, tokens.data)
|
|
|
|
scores = self.model.score(context)
|
2015-01-17 05:21:17 +00:00
|
|
|
tokens.data[i].tag = arg_max(scores, self.model.n_classes)
|
2014-12-31 08:40:59 +00:00
|
|
|
self.set_morph(i, tokens.data)
|
2015-01-24 06:29:04 +00:00
|
|
|
tokens._tag_strings = self.tag_names
|
2014-12-21 09:59:07 +00:00
|
|
|
|
2014-12-31 08:40:59 +00:00
|
|
|
def train(self, Tokens tokens, object golds):
|
2014-12-21 09:59:07 +00:00
|
|
|
cdef int i
|
|
|
|
cdef atom_t[N_CONTEXT_FIELDS] context
|
2014-12-31 08:40:59 +00:00
|
|
|
cdef const weight_t* scores
|
2014-12-30 10:20:15 +00:00
|
|
|
correct = 0
|
2014-12-21 09:59:07 +00:00
|
|
|
for i in range(tokens.length):
|
2014-12-31 08:40:59 +00:00
|
|
|
fill_context(context, i, tokens.data)
|
|
|
|
scores = self.model.score(context)
|
|
|
|
guess = arg_max(scores, self.model.n_classes)
|
|
|
|
self.model.update(context, guess, golds[i], guess != golds[i])
|
2015-01-17 05:21:17 +00:00
|
|
|
tokens.data[i].tag = guess
|
2014-12-31 08:40:59 +00:00
|
|
|
self.set_morph(i, tokens.data)
|
|
|
|
correct += guess == golds[i]
|
2014-12-30 10:20:15 +00:00
|
|
|
return correct
|
2014-12-21 09:59:07 +00:00
|
|
|
|
2014-12-23 02:18:59 +00:00
|
|
|
cdef int set_morph(self, const int i, TokenC* tokens) except -1:
|
2015-01-17 05:21:17 +00:00
|
|
|
cdef const PosTag* tag = &self.tags[tokens[i].tag]
|
2014-12-24 06:42:00 +00:00
|
|
|
tokens[i].pos = tag.pos
|
2015-01-22 15:08:25 +00:00
|
|
|
cached = <_CachedMorph*>self._morph_cache.get(tag.id, tokens[i].lex.orth)
|
2014-12-23 02:18:59 +00:00
|
|
|
if cached is NULL:
|
|
|
|
cached = <_CachedMorph*>self.mem.alloc(1, sizeof(_CachedMorph))
|
|
|
|
cached.lemma = self.lemmatize(tag.pos, tokens[i].lex)
|
|
|
|
cached.morph = tag.morph
|
2015-01-22 15:08:25 +00:00
|
|
|
self._morph_cache.set(tag.id, tokens[i].lex.orth, <void*>cached)
|
2014-12-23 02:18:59 +00:00
|
|
|
tokens[i].lemma = cached.lemma
|
|
|
|
tokens[i].morph = cached.morph
|
|
|
|
|
2015-01-11 23:26:22 +00:00
|
|
|
cdef int lemmatize(self, const univ_tag_t pos, const LexemeC* lex) except -1:
|
2014-12-23 02:18:59 +00:00
|
|
|
if self.lemmatizer is None:
|
2015-01-22 15:08:25 +00:00
|
|
|
return lex.orth
|
|
|
|
cdef unicode py_string = self.strings[lex.orth]
|
2014-12-23 02:18:59 +00:00
|
|
|
if pos != NOUN and pos != VERB and pos != ADJ:
|
2015-01-22 15:08:25 +00:00
|
|
|
return lex.orth
|
2014-12-23 02:18:59 +00:00
|
|
|
cdef set lemma_strings
|
2015-01-05 00:54:29 +00:00
|
|
|
cdef unicode lemma_string
|
2015-01-13 13:03:48 +00:00
|
|
|
lemma_strings = self.lemmatizer(py_string, pos)
|
2014-12-23 02:18:59 +00:00
|
|
|
lemma_string = sorted(lemma_strings)[0]
|
2015-01-05 00:54:29 +00:00
|
|
|
lemma = self.strings.intern(lemma_string.encode('utf8'), len(lemma_string)).i
|
2014-12-23 02:18:59 +00:00
|
|
|
return lemma
|
|
|
|
|
|
|
|
def load_morph_exceptions(self, dict exc):
|
|
|
|
cdef unicode pos_str
|
|
|
|
cdef unicode form_str
|
|
|
|
cdef unicode lemma_str
|
|
|
|
cdef dict entries
|
|
|
|
cdef dict props
|
|
|
|
cdef int lemma
|
2015-01-22 15:08:25 +00:00
|
|
|
cdef id_t orth
|
2014-12-23 02:18:59 +00:00
|
|
|
cdef int pos
|
|
|
|
for pos_str, entries in exc.items():
|
|
|
|
pos = self.tag_names.index(pos_str)
|
|
|
|
for form_str, props in entries.items():
|
|
|
|
lemma_str = props.get('L', form_str)
|
2015-01-22 15:08:25 +00:00
|
|
|
orth = self.strings[form_str]
|
2014-12-23 02:18:59 +00:00
|
|
|
cached = <_CachedMorph*>self.mem.alloc(1, sizeof(_CachedMorph))
|
|
|
|
cached.lemma = self.strings[lemma_str]
|
|
|
|
set_morph_from_dict(&cached.morph, props)
|
2015-01-22 15:08:25 +00:00
|
|
|
self._morph_cache.set(pos, orth, <void*>cached)
|
2014-12-23 02:18:59 +00:00
|
|
|
|
2014-12-21 09:59:07 +00:00
|
|
|
|
|
|
|
cdef int fill_context(atom_t* context, const int i, const TokenC* tokens) except -1:
|
2015-01-22 15:08:25 +00:00
|
|
|
_fill_from_token(&context[P2_orth], &tokens[i-2])
|
|
|
|
_fill_from_token(&context[P1_orth], &tokens[i-1])
|
|
|
|
_fill_from_token(&context[W_orth], &tokens[i])
|
|
|
|
_fill_from_token(&context[N1_orth], &tokens[i+1])
|
|
|
|
_fill_from_token(&context[N2_orth], &tokens[i+2])
|
2014-12-21 09:59:07 +00:00
|
|
|
|
|
|
|
|
|
|
|
cdef inline void _fill_from_token(atom_t* context, const TokenC* t) nogil:
|
2015-01-22 15:08:25 +00:00
|
|
|
context[0] = t.lex.orth
|
2014-12-21 09:59:07 +00:00
|
|
|
context[1] = t.lex.cluster
|
|
|
|
context[2] = t.lex.shape
|
|
|
|
context[3] = t.lex.prefix
|
|
|
|
context[4] = t.lex.suffix
|
2015-01-24 15:20:15 +00:00
|
|
|
context[5] = t.tag
|
2014-12-21 09:59:07 +00:00
|
|
|
context[6] = t.lemma
|