def make_vocab_files():
"""
Produce the question and answer vocabulary files.
"""
print 'making question vocab...', config.QUESTION_VOCAB_SPACE
qdic, _ = VQADataProvider.load_data(config.QUESTION_VOCAB_SPACE)
question_vocab = make_question_vocab(qdic)
print 'making answer vocab...', config.ANSWER_VOCAB_SPACE
_, adic = VQADataProvider.load_data(config.ANSWER_VOCAB_SPACE)
answer_vocab = make_answer_vocab(adic, config.NUM_OUTPUT_UNITS)
return question_vocab, answer_vocab
def make_question_vocab(qdic):
"""
Returns a dictionary that maps words to indices.
"""
vdict = {'':0}
vid = 1
for qid in qdic.keys():
# sequence to list
q_str = qdic[qid]['qstr']
q_list = VQADataProvider.seq_to_list(q_str)
# create dict
for w in q_list:
if not vdict.has_key(w):
vdict[w] = vid
vid +=1
return vdict