def memm_greedy(sentences, max_sentence_len, features_map, counters_dict, clf):
all_words_features = list()
sentences_predictions = list()
for j in range(len(sentences)):
sentences_predictions.append(list())
for i in range(max_sentence_len):
start = datetime.now()
for j in range(len(sentences)):
sentence = sentences[j]
if i < len(sentence):
word = sentence[i]
word_features = FeaturesUtils.get_word_features(
i, sentence, sentences_predictions[j],
DictUtils.is_rare(counters_dict, word))
all_words_features.append(word_features)
else:
all_words_features.append(dict())
all_prediction_word_i = get_prediction_of_all_words(
all_words_features, clf, features_map)
end = datetime.now()
print('word i={1} Running Time: {0}'.format(end - start, i))
for j in range(len(all_prediction_word_i)):
sentences_predictions[j].append(all_prediction_word_i[j])
all_words_features = list()
return sentences_predictions
def create_features(words_features_list, words, tags, dict_e):
for i in range(len(words)):
is_rare = DictUtils.is_rare(dict_e, (words[i], tags[i]))
word_feature_dict = FeaturesUtils.get_word_features(
i, words, tags, is_rare)
word_feature_dict['tag'] = tags[i]
words_features_list.append(word_feature_dict)
print(word_feature_dict)
def get_word_features_list(i, words, prev_predictions, prev_prev_predictions,
counters_dict):
all_word_features = list()
prev_list = list()
for prev_prediction in prev_predictions:
for prev_prev_prediction in prev_prev_predictions:
word_features = FeaturesUtils.get_word_features(
i, words, get_tag_list(i, prev_prediction,
prev_prev_prediction),
DictUtils.is_rare(counters_dict, words[i]))
all_word_features.append(word_features)
prev_list.append((prev_prediction, prev_prev_prediction))
return all_word_features, prev_list
