def post(self):
K.clear_session()
model, graph = init_model()
hashtag = session.get('hashtag')
tokenized = session.get('tokenized')
# for table display
tweets = TweetsModel.find_by_hashtag(hashtag=hashtag)
tweets = [element.tweet for element in tweets]
# apply model
predictions = []
with graph.as_default():
for element in tokenized:
element = pad_sequences(element, maxlen=23, dtype='int32', value=0)
prediction = model.predict(element, batch_size=1, verbose=2)[0]
if (np.argmax(prediction) == 0):
prediction = 'Negative'
elif (np.argmax(prediction) == 1):
prediction = 'Neutral'
else:
prediction = 'Positive'
predictions.append(prediction)
# zip predictions with ORIGINAL tweet for better clarity
sentiment_predictions = list(zip(tweets, predictions))
sentiment_counter = dict(Counter(predictions))
return {'sentiment_predictions': sentiment_predictions, 'sentiment_counter': sentiment_counter,
'hashtag': hashtag}
def manage_db():
"""Manage Databases View"""
mongo_hashtags = Collection.hashtags()
sqlite_hashtags = TweetsModel.distinct_hashtags()
return render_template('manage_db/dashboard.html',
mongo_hashtags=mongo_hashtags,
sqlite_hashtags=sqlite_hashtags)
def manage_db():
"""Manage Databases View"""
for attempt in range(0, 3):
try:
mongo_hashtags = Collection.hashtags()
except AutoReconnect:
time.sleep(2)
sqlite_hashtags = TweetsModel.distinct_hashtags()
return render_template('manage_db/dashboard.html',
mongo_hashtags=mongo_hashtags,
sqlite_hashtags=sqlite_hashtags)
def post(self):
hashtag = request.form.get('hashtag_cleaning')
# Prevent overpopulating
TweetsModel.delete_by_hashtag(hashtag=hashtag)
if hashtag is None:
return redirect('trendsearch', message="hashtag was not specified")
# Get chosen hashtag tweets
hashtag_tweets = Collection.find_by_hashtag(hashtag=hashtag)
# Preprocess text for future senitment analysis
text = [element['text'] for element in hashtag_tweets]
corpus = []
for i in range(0, len(text)):
try:
tweet = re.sub('[^a-zA-Z]', ' ', text[i]) ## all the indexes
tweet = tweet.lower()
tweet = tweet.split()
ps = PorterStemmer()
tweet = [
ps.stem(word) for word in tweet if not word in stopwords
]
tweet = ' '.join(tweet)
corpus.append(tweet)
new_tweet = TweetsModel(hashtag=hashtag, tweet=tweet)
new_tweet.save_to_db()
except Exception as e:
pass
return redirect('manage_db')
def post(self):
hashtag = request.form.get('hashtag')
# for backend sentiment analysis
hashtag_tweets = TweetsModel.find_by_hashtag(hashtag=hashtag)
hashtag_tweets = [element.tweet for element in hashtag_tweets]
# load pickled tokenizer
with open(tokenizer_path, 'rb') as handle:
tokenizer = pickle.load(handle)
# preprocess data for prediction
tokenized = []
for element in hashtag_tweets:
element = element.split()
element = tokenizer.texts_to_sequences(element)
tokenized.append(element)
session['tokenized'] = tokenized
session['hashtag'] = hashtag
return hashtag
def sentiment_analysis():
"""Apply LSTM View"""
sqlite_hashtags = TweetsModel.distinct_hashtags()
return render_template('sentiment_analysis/dashboard.html',
sqlite_hashtags=sqlite_hashtags)
def post(self):
hashtag = request.form.get('hashtag_table')
TweetsModel.delete_by_hashtag(hashtag=hashtag)
return redirect('manage_db')