Mean Tweet Reader

Inspiration

We were given a "Hack Harassment" card and we wanted to find a way to track negative online interactions.

What it does

Our hack collects recent tweets from a user's twitter account and uses a trained natural language processor to calculate a "harassment" rating.

How we built it

We used Python to implement a natural language processor to determine negative and potentially abusive tweets. The twitter data was collected from a Twitter API python library and manually assigned "good" and "bad" ratings.

Challenges we ran into

It was difficult to parse clean data and testing natural language libraries that we could use.Teaching the language processor takes a large amount of data and manpower that is hard to manage within time constraints.

Accomplishments that we're proud of

We got relatively accurate ratings of mean tester accounts! (these accounts were actually temporarily banned for abusive behavior).

What we learned

Dealing with large data is difficult and collecting relevant data takes time.

What's next for Mean Tweet Reader

More accurate ratings by teaching our natural language processor with more data. We could also teach it to deal with images or emojis and create a anti-harassment mechanism for abusive users.

Built With

• appjar
• python