A hackathon project from #peacehackldn

Seeking to turn hateful tweets into something a little more funny

From

to

See -h flag for running instructions.

Sentiment analysis is done for an account to find most hateful tweet scoring using naive Bayes classifier and a formula we devised taking into account ./dict/terrible.txt words, using words found under ./dict/bad... and ./dict/good... replacement is done.

• Better replacement and training data [Hatebase]

Find someone's influencing more friendly friends, via graph representation of a twitter account and their followers with colour coding according to sentiment analysis score

Averages our sentiment score from their most recent n tweets and from p followers and their followers _{and their followers and their...}, repeated q times

• Filter according to interests so you can find
• Hover show twitter handles

• nltk plus punkt and either movie_reviews or whatever corpus you wanna train on
• twython

This code would not be possible without code shared from streamhacker for NB classifier, networkx for their json graph deployment and heatmap for colours and layout.