def get_word_signatures_tag(word, dict_e, unk_tag_list):
signatures = WordSignatures.get_word_signatures(word)
if signatures == [word.lower()]:
return {UNK: unk_tag_list}
else:
signatures_tags = dict()
for signature in signatures:
signatures_tags[signature] = DictUtils.possible_tags(signature, dict_e)
return signatures_tags
def main(input_file_name, q_mle, e_mle, greedy_hmm_output, extra_file_name):
start = datetime.now()
sentences = FileUtils.read_lines(input_file_name)
dict_q = DictUtils.convert_line_to_dict(FileUtils.read_lines(q_mle))
dict_e = DictUtils.convert_line_to_dict(FileUtils.read_lines(e_mle))
unk_tag_list = DictUtils.possible_tags(UNK, dict_e)
tagged_text = greedy(sentences, dict_q, dict_e, unk_tag_list)
FileUtils.write_tagged_text(greedy_hmm_output, tagged_text)
end = datetime.now()
print('Running Time: {0}'.format(end - start))
def main(input_file_name, q_mle, e_mle, hmm_viterbi_predictions, extra_file_name):
start = datetime.now()
sentences = FileUtils.read_lines(input_file_name)
dict_q = DictUtils.convert_line_to_dict(FileUtils.read_lines(q_mle))
dict_e = DictUtils.convert_line_to_dict(FileUtils.read_lines(e_mle))
unk_tag_list = DictUtils.possible_tags('*UNK*', dict_e)
tagged_text = viterbi(sentences, dict_q, dict_e, unk_tag_list)
FileUtils.write_tagged_text(hmm_viterbi_predictions, tagged_text)
end = datetime.now()
print('Running Time: {0}'.format(end - start))
def possible_tags(word, dict_e, unk_tag_list):
words_tags = DictUtils.possible_tags(word, dict_e)
if len(words_tags) == 0:
return get_word_signatures_tag(word, dict_e, unk_tag_list)
else:
return {word: words_tags}
