def run_test(text_sentences, sentences, sentence_pos, trans_probs, em_probs, proto_index, out_file, dictionary=None):
out_text_sents = []
correct = 0
total = 0
s = 0
confusion_matrix = numpy.zeros((n_pos, n_pos))
for sentence, sent_pos in zip(sentences, sentence_pos):
pred_pos = viterbi(sentence, trans_probs, em_probs, dictionary)
w = 0
out_text_sent = []
for pred, gold, word in zip(pred_pos, sent_pos, sentence):
total += 1
if pred == gold:
correct += 1
out_text_sent.append((text_sentences[s][w], pos_tags[pred]))
w += 1
confusion_matrix[pred, gold] += 1
out_text_sents.append(out_text_sent)
s += 1
print '------------------'
print 'accuracy:', correct / float(total)
print 'many to one:', confusion_matrix.max(axis=1).sum() / float(confusion_matrix.sum())
print '------------------'
if out_file is not None:
load_and_save.write_sentences_to_file(out_text_sents, out_file)
def run_test(text_sentences,
sentences,
sentence_pos,
trans_probs,
em_probs,
proto_index,
out_file,
dictionary=None):
out_text_sents = []
correct = 0
total = 0
s = 0
confusion_matrix = numpy.zeros((n_pos, n_pos))
for sentence, sent_pos in zip(sentences, sentence_pos):
pred_pos = viterbi(sentence, trans_probs, em_probs, dictionary)
w = 0
out_text_sent = []
for pred, gold, word in zip(pred_pos, sent_pos, sentence):
total += 1
if pred == gold:
correct += 1
out_text_sent.append((text_sentences[s][w], pos_tags[pred]))
w += 1
confusion_matrix[pred, gold] += 1
out_text_sents.append(out_text_sent)
s += 1
print '------------------'
print 'accuracy:', correct / float(total)
print 'many to one:', confusion_matrix.max(axis=1).sum() / float(
confusion_matrix.sum())
print '------------------'
if out_file is not None:
load_and_save.write_sentences_to_file(out_text_sents, out_file)