def tag(self, test_data_filename, result_filename):
# prepare output
output = file(result_filename, 'w')
# calcuate arg max e('_RARE_'|y)
max_e_rare_y, max_rare_y = -1.0, ''
for tag in self.tags:
e = self.emission_params[(util.RARE_TAG, tag)]
if e > max_e_rare_y:
max_e_rare_y, max_rare_y = e, tag
# load test_data
word_iterator = util.test_data_iterator(file(test_data_filename))
# tag
for word in word_iterator:
max_e, max_y = -1.0, ''
if word is not None:
if word == util.RARE_TAG or word not in self.words:
max_y = max_rare_y
else:
for tag in self.tags:
e = self.emission_params[(word, tag)]
if e > max_e:
max_e, max_y = e, tag
output.write('{0} {1}\n'.format(word, max_y))
else:
output.write('\n')
def tag(self, test_data_filename, result_filename):
# Preparando la salida
output = file(result_filename, 'w')
# calculando el maximo arg max e('_RARE_'|y)
max_e_rare_y, max_rare_y = -1.0, ''
for tag in self.tags:
e = self.emission_params[(util.RARE_TAG, tag)]
if e > max_e_rare_y:
max_e_rare_y, max_rare_y = e, tag
# cargando el data test
word_iterator = util.test_data_iterator(file(test_data_filename))
# etiquetado
for word in word_iterator:
max_e, max_y = -1.0, ''
if word is not None:
if word == util.RARE_TAG or word not in self.words:
max_y = max_rare_y
else:
for tag in self.tags:
e = self.emission_params[(word, tag)]
if e > max_e:
max_e, max_y = e, tag
output.write('{0} {1}\n'.format(word, max_y))
else:
output.write('\n')
def tag(self, test_data_file, result_file):
sent_iterator = util.test_sent_iterator(
util.test_data_iterator(test_data_file))
for sent in sent_iterator:
tags = self.viterbi(sent)
for s, t in zip(sent, tags):
result_file.write('{0} {1}\n'.format(s, t))
result_file.write('\n')
def tag(self, test_data_file, result_file):
# get test sentence
sent_iterator = util.test_sent_iterator(
util.test_data_iterator(test_data_file))
for sent in sent_iterator:
tags = self.viterbi(sent)
for s, t in zip(sent, tags):
result_file.write('{0} {1}\n'.format(s, t))
result_file.write('\n')
def tag(self, test_data_file, result_file):
# Obteniendo las oraciones de prueba
sent_iterator = util.test_sent_iterator(
util.test_data_iterator(test_data_file))
for sent in sent_iterator:
tags = self.viterbi(sent)
for s, t in zip(sent, tags):
result_file.write('{0} {1}\n'.format(s, t))
result_file.write('\n')
