Web app which analyzes the sentiment of tweets of a specific theme (Portuguese only).
The dataset used for the training of the classifier was scrapped from tripadvior restaurants' reviews.
The classifier is saved in a .pkl file. It is the svm3class.pkl. It categorizes the tweets in POSITIVE, NEGATIVE and NEUTRAL.
svm2class.pkl it is not used, but it can classifiy tweets between POSITIVE and NEGATIVE only.
https://twisentiment.herokuapp.com/
Create a virtual environment and export variables, but first you need to go to Twitter api and get tokens and keys.
export consumer_key=CONSUMER_KEY
export consumer_token=CONSUMER_TOKEN
export access_token=ACCESS_TOKEN
export access_secret=ACCESS_SECRET
run python app.py