Bot or Not

Twitter bots can be a problem. Disinformation and spam are just some of the problems they cause. This project takes a twitter user ID, and uses a Neural Net to determine if they may be a bot.

Comment

(X will be referred to as Twitter for this post)

Inspiration

Wanting to deal with over saturation of bot content and intentionally rage inciting accounts on social media, specifically Twitter.

What it does

A user enters a Twitter user ID and is returned a number between 0 and 1, indicating the probability that the account is a bot or a real person. 1 being 100% a bot, 0 being 0%.

How we built it

First, we scoured Twitter (yikes) to find accounts that we knew were real, then looked for accounts that we were pretty sure after looking at many of their posts that they were bots (using some of the criteria listed on our website). Next, Chris did the development of the backend that would analyze an account using a Neural Net he would develop. At the same time, team members Jacob and Steven were trying to figure out javascript for the first time. Jacob stared at javascript for a few hours and proceeded to go do work researching methods of identifying bots, finding some bot accounts, and making the presentation, because he gave up on javascript. Steven, having gotten somewhat of a grasp after also pounding his head against a wall for hours, was able to start designing the webpage.

Challenges we ran into

The two people on our team assigned to the front end of the website had never done webdev before, and also never touched javascript.

Accomplishments that we're proud of

Steven, one of the team members that has never done webdev, was able to get the UI for the website put together and looking pretty sleek. Chris was able to get the Neural Net we use to analyze the validity of whether an account is a bot or bot working!

What we learned

For future hackathons, knowing webdev is really important unless you're planning to make something like a game.

What's next for Bot or Not

The website UI shouldn't need to change much, it's basic functionality of taking in a Twitter user ID and outputting a percentage may stay how it is, but the future changes would be to the analysis the Neural Net does. More information would be collected from each individual account, such as looking at all of the posts made by an account to do sentiment analysis, or determing if there are mistakes in the language of posts that would indicate the post was not originally written in the language it was posted in.

Built With

Share this project:

Updates