spaCy/spacy/tagger.pyx

142 lines
4.4 KiB
Cython

# cython: profile=True
from __future__ import unicode_literals
from __future__ import division
from os import path
import os
import shutil
import random
import json
import cython
from thinc.features cimport Feature, count_feats
def setup_model_dir(tag_names, tag_map, tag_counts, 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,
'tag_counts': tag_counts,
}
with open(path.join(model_dir, 'config.json'), 'w') as file_:
json.dump(config, file_)
cdef class Tagger:
"""Predict some type of tag, using greedy decoding. The tagger reads its
model and configuration from disk.
"""
def __init__(self, model_dir):
self.mem = Pool()
cfg = json.load(open(path.join(model_dir, 'config.json')))
templates = cfg['templates']
tag_map = cfg['tag_map']
univ_counts = {}
cdef unicode tag
cdef unicode univ_tag
self.tag_names = cfg['tag_names']
self.tags = <PosTag*>self.mem.alloc(len(self.tag_names), sizeof(PosTag))
for i, tag in enumerate(self.tag_names):
pos, props = tag_map[tag]
self.tags[i].id = i
self.tags[i].pos = pos
self.tags[i].morph.number = props.get('number', 0)
self.tags[i].morph.tenspect = props.get('tenspect', 0)
self.tags[i].morph.mood = props.get('mood', 0)
self.tags[i].morph.gender = props.get('gender', 0)
self.tags[i].morph.person = props.get('person', 0)
self.tags[i].morph.case = props.get('case', 0)
self.tags[i].morph.misc = props.get('misc', 0)
self.tagdict = _make_tag_dict(cfg['tag_counts'])
self.extractor = Extractor(templates)
self.model = LinearModel(len(self.tag_names), self.extractor.n_templ+2)
if path.exists(path.join(model_dir, 'model')):
self.model.load(path.join(model_dir, 'model'))
cdef class_t predict(self, atom_t* context, object golds=None) except *:
"""Predict the tag of tokens[i].
>>> tokens = EN.tokenize(u'An example sentence.')
>>> tag = EN.pos_tagger.predict(0, tokens)
>>> assert tag == EN.pos_tagger.tag_id('DT') == 5
"""
cdef int n_feats
cdef Feature* feats = self.extractor.get_feats(context, &n_feats)
cdef weight_t* scores = self.model.get_scores(feats, n_feats)
guess = _arg_max(scores, self.model.nr_class)
if golds is not None and guess not in golds:
best = _arg_max_among(scores, golds)
counts = {guess: {}, best: {}}
count_feats(counts[guess], feats, n_feats, -1)
count_feats(counts[best], feats, n_feats, 1)
self.model.update(counts)
return guess
def tag_id(self, object tag_name):
"""Encode tag_name into a tag ID integer."""
tag_id = self.tag_names.index(tag_name)
if tag_id == -1:
tag_id = len(self.tag_names)
self.tag_names.append(tag_name)
return tag_id
UNIV_TAGS = {
'NULL': NO_TAG,
'ADJ': ADJ,
'ADV': ADV,
'ADP': ADP,
'CONJ': CONJ,
'DET': DET,
'NOUN': NOUN,
'NUM': NUM,
'PRON': PRON,
'PRT': PRT,
'VERB': VERB,
'X': X,
'.': PUNCT,
'EOL': EOL
}
def _make_tag_dict(counts):
freq_thresh = 50
ambiguity_thresh = 0.98
tagdict = {}
cdef atom_t word
cdef atom_t tag
for word_str, tag_freqs in counts.items():
tag_str, mode = max(tag_freqs.items(), key=lambda item: item[1])
n = sum(tag_freqs.values())
word = int(word_str)
tag = int(tag_str)
if n >= freq_thresh and (float(mode) / n) >= ambiguity_thresh:
tagdict[word] = tag
return tagdict
cdef class_t _arg_max(weight_t* scores, int n_classes) except 9000:
cdef int best = 0
cdef weight_t score = scores[best]
cdef int i
for i in range(1, n_classes):
if scores[i] >= score:
score = scores[i]
best = i
return best
cdef class_t _arg_max_among(weight_t* scores, list classes):
cdef int best = classes[0]
cdef weight_t score = scores[best]
cdef class_t clas
for clas in classes:
if scores[clas] > score:
score = scores[clas]
best = clas
return best