def diary(request):
if request.method == "POST":
diary = request.POST.get("diary")
usern = request.POST.get("name")
print(diary)
try:
print('predict----->', predict(diary))
sentim = predict(diary)
except:
pass
sentimentp = TextBlob(diary).sentiment.polarity
if(sentimentp < 0):
sentimentp = -1
elif(sentimentp > 0):
sentimentp = 1
else:
sentimentp = 0
try:
usern = UserModel.objects.get(id=usern)
userdiary = DiaryModel.objects.create(
description=diary, polarity=sentimentp, user=usern)
userdiary.save()
return redirect("home")
except:
return HttpResponse("failed")
else:
return render(request, 'diary.html')
def POST(self, name):
logging.debug(web.data())
text = web.input(text="").text
prediction_result = sentiment.predict(text)
render = web.template.render('templates/')
return render.index(prediction_result)
def GET(self, name):
logging.debug(web.data())
sentence = web.input(sentence="").sentence
logging.debug('sentence: ' + sentence)
start_time = current_milli_time()
result = sentiment.predict(sentence)
res = {
'status': 'ok',
'took': current_milli_time() - start_time,
'result': str(result)
}
web.header('Content-Type', 'application/json')
return json.dumps(res)
def callback(ch, method, properties, body):
tweets = json.loads(body.decode())
for t in tweets['tweets']:
pred = predict(_deemojify(t['tweet']))
t['sentiment'] = 1 if pred else 0
if len(tweets['tweets']) > 0:
sent_perc = sum([t["sentiment"] for t in tweets["tweets"]
]) / len(tweets["tweets"])
else:
sent_perc = 0
logger.info(
f'{tweets["id_hackadeira"][:8]} :: {100*sent_perc:.2f}%'
) # noqa
ch.basic_ack(delivery_tag=method.delivery_tag)
_send_to_queue(tweets)
def ask():
# kernel now ready for use
while True:
message = str(request.form['messageText'])
m1 = message[:5]
if message == "quit":
aimodule.record()
elif message == "save":
aimodule.saveBrain("bot_brain.brn")
elif message == "#elp":
aimodule.incident()
else:
bot_response = aimodule.respond(message)
sen_response = sentiment.predict(message)
resp = bot_response + "---" + sen_response
#print sen_response
logger.debug(sen_response)
return jsonify({'status': 'OK', 'answer': resp})
def GET(self, name):
logging.debug(web.data())
search = web.input(search="").search
logging.debug('sentence: ' + search)
start_time = current_milli_time()
result = {}
twitter_search = twitterClient.search(search)
for idx, twit in enumerate(twitter_search):
result[str(idx)] = {'text': twit, 'rate': sentiment.predict(twit)}
res = {
'status': 'ok',
'took': current_milli_time() - start_time,
'result': result
}
web.header('Content-Type', 'application/json')
return json.dumps(res,
sort_keys=True,
indent=4,
separators=(',', ': '))
def calculate_would_buy():
friendships, reviews = import_data()
clusters, scores, purchased = import_results()
with open('model.pkl', 'rb') as f:
model = pickle.load(f)
vocab_index = model['vocabulary_index']
term_probability_matrix = model['term_probability_matrix']
class_prob = model['class_probability']
with open('community_detection/communities_test.p', 'rb') as f:
communities = pickle.load(f)
logger.debug('Loaded data')
# Replaces the review with a label.
for user, review in reviews.items():
data = preprocessing(review)
reviews[user] = predict(data, vocab_index, term_probability_matrix,
class_prob)
logger.debug('Predicted reviews')
would_purchase = calculate_answer(communities, friendships, reviews)
logger.debug('Calculated would purchase')
scores = convert_to_balanced(scores)
print_cluster_yes_percentage(communities, would_purchase)
print_cluster_accuracy(communities, clusters)
print_review_accuracy(reviews, scores)
print_review_precision(reviews, scores, 1)
print_review_recall(reviews, scores, 1)
print_review_precision(reviews, scores, 0)
print_review_recall(reviews, scores, 0)
print_purchase_accuracy(would_purchase, purchased)
with open('would_purchase.pkl', 'wb') as f:
pickle.dump(would_purchase, f)
def test(self):
s = "Dear Alex,\n\nMahalo. The movie is awesome. I didn't ruin it. \n\nEnjoy,\n\nD to the J https://t.co/rDyir8hgzi"
prediction = sentiment.predict(s)
print(prediction)
self.assertGreater(prediction['pos'], prediction['neg'])