def training_ignor_class(request, tid):
t = GroupTopic.objects.filter(tid=tid).first()
st = time.time()
sample = get_topicatbs(t)
nbc = NaiveBayesClassifier(Drive())
r = nbc.classify(sample)
res = '%s %s' % (r, time.time() - st)
return HttpResponse(res)
def findad(request, tid=''):
t = None
if tid:
t = GroupTopic.objects.filter(tid=tid, other='').first()
else:
t = GroupTopic.objects.filter(type='', other='').first()
if not t:
return new_topic(request)
st = time.time()
sample = get_topicatbs(t)
nbc = NaiveBayesClassifier(Drive())
r = nbc.classify(sample)
ct = time.time() - st
try:
t.type = max(r.items(), key=lambda x:x[1])[0]
if 'ad' in r and 'ok' in r:
p = r['ad'] / (r['ad'] + r['ok'])
t.other = 'auto'
st = time.time()
if p > 0.60:
# 广告
t.other = t.other + " ad"
try:
# comment_topic(t.tid, 'mark %s' % p)
report_topic(t.tid)
except:
t.other = 'failed'
if 'po' in r and 'ok' in r:
p = r['po'] / (r['po'] + r['ok'])
if p > 0.60:
# 色情
t.other = t.other + " po"
try:
report_topic(t.tid, '1')
except:
t.other = 'failed'
except:
t.other = 'failed'
t.save()
rt = time.time() - st
res = {
'title':t.title,
'url':t.url,
'tid':t.tid,
'p':p,
'type':t.type,
'other':t.other,
'class_time':ct,
'report_time':rt
}
return HttpResponse(json.dumps(res))
def training_to_class(request, tid, cls):
t = GroupTopic.objects.filter(tid=tid).first()
st = time.time()
res = None
if t.other == 'train':
res = 'trained'
else:
sample = get_topicatbs(t)
nbc = NaiveBayesClassifier(Drive())
nbc.training(sample, cls)
t.other = 'train'
t.type = cls
t.save()
res = '%s %s' % ('/'.join(sample), time.time() - st)
return HttpResponse(res)
def tdm(request, txt):
st = time.time()
nbc = NaiveBayesClassifier(Drive())
# test_classifer(nbc)
c = nbc.whichclass(txt)
return HttpResponse('%ss %s'%(time.time()-st, c))
