HaterDetector

• 

  Context for a Tweet

• 

  Tweets for First run

• 

  Some more tweets

Inspiration

There are tonne of people online who want to harass people, just for the sake of it. Obviously, it is not socially acceptable and while we cannot stop everyone, we can at least call them out for it when no one does.

What it does

It's a simple twitter account that we are creating which will be run by a bot. The bot will identify people who are rude and mean to others online and call them out for it.

How we built it

We used Twitter API and my already existing open source project for social media mining to build a twitter bot that will identify people by filtering them with the use of common words haters use online. In its initial phase, the system will first collect data from manual keywords. We will then run it through a Natural Language Processor to understand the semantics of it. Once we are done with that, we will then run our bot as an automated script and it will simply call out people for being mean by posting a meme, etc. to them.

Challenges we ran into

Our Twitter Bot originally created was blocked by Twitter Systems. We overcame that problem by using one of our team member's account for this bot. Another issue was that we were not allowed the free-tier for AWS which was a crucial part of our project. So what we did was we use our own money and got the paid tier to work with AWS.

Accomplishments that we're proud of

Achieving so much in so less time, and overcoming many hurdles, working hard together all night and most importantly creating something meaningful

What we learned

We learned that once we get our minds to it, we can get anything done. Some of the team members were not familiar with the technology stack used for this project as much as others but we all picked up with the pace and contributed to making an amazing project which we actually hope to open source later.

What's next for HaterDetector

We plan to open source out the project, explore more machine learning algorithms to identify ones that will be more cost efficient and accurate and then lease it out and extend to other online communities as well.

Built With

• amazon-web-services
• git
• machine-learning
• natural-language-processing
• python
• tweepy
• twitter