def serv(model):
decoder = NaiveDecoder(model)
while True:
query = raw_input('Input your query(must be segmented by SPACE), q to quit:\n').decode('utf-8')
if query == u'q':
return
domains = raw_input('Input the domains you want to compare:\n').decode('utf-8')
if not domains:
domains = decoder.model.domains
else:
domains = domains.split(' ')
query = extract(query)
ret = decoder.predict(query)
print "\n%s\n%s\n" % (ret, '=' * 50)
lst = []
for domain in domains:
score, detail = decoder.get_score(query.split(' '), domain)
lst.append((score, domain, detail))
lst.sort(key = lambda x: -x[0])
for domain, score, detail in lst:
print score, domain
for term in query.split(' '):
cate = decoder.model.get_category(term)
sys.stdout.write('%s(%s, freq:%d, gini:%.3f): %.4f\t' % \
(term, cate, decoder.model.term_count[cate],
get_gini(cate), detail[cate][1]))
print '\n%s\n' % ('-' * 20)
def extract(sent):
return ' '.join(sorted(list(set(sent.split(' '))),
key = lambda term: -get_gini(term_category(term)))[:5])