def create_tagged_file(emission_params,filename,freq_words):
read_file = open(filename,'rU')
write_file_path = get_write_file_path(filename,'.tagged.classes')
write_file = open(write_file_path, 'w')
for line in read_file:
if line.strip() == '':
write_file.write('\n')
continue
word = line.strip()
tag = get_tag(word,emission_params,freq_words)
write_file.write(word + ' ' + tag + '\n')
read_file.close()
write_file.close()
def create_viterbi(emission_params,filename,freq_words):
read_file = open(filename,'rU')
write_file_path = get_write_file_path(filename,'.viterbi.classes')
write_file = open(write_file_path, 'w')
sentence = []
for line in read_file:
if line.strip() == '':
sentence.append('STOP')
viterbi_tags = viterbi(emission_params,freq_words,sentence)
write_viterbi_tags(sentence,viterbi_tags,write_file)
sentence = []
else:
word = line.strip()
sentence.append(word)
read_file.close()
write_file.close()