def prepare_entry(rv, date=None, rvid=None, rating=None):
entry = dict()
# start = time.time()
if date is not None:
entry["date"] = date
if rvid is not None:
entry["review_id"] = rvid
if rating is not None:
entry["review_score"] = rating
entry['mention'] = rv
question_toks = CSV.process_nl(rv)
entry['input_tokens'] = question_toks
# print("- TIMING: %.4f seconds for NL tokenization" % (time.time() - start))
return entry
def prepare_nls(nls):
data = []
for nl in nls:
data.append((CSV.process_nl(nl), None))
return data