Inspiration
The inspiration behind Misbehavin' is to help identify potential bullies on Twitter. It can be difficult for users to manually report posts that engage in hate speech, so with the help of the expert.ai NLP API, we can automate this process and identify accounts that may be in violation of not only Twitter's Terms of Service agreement, but also to identify those who just lack common decency and compassion. It's a very simple app -- all you need is to know the Twitter handle of an account that you would like to check.
What it does
Misbehavin' uses the Twitter API to aggregate a list of the most recent 10 tweets from any user. It then runs these tweets through the expert.ai hate speech detector, which will let us know if any of the Tweets fit the following categories: Personal Insult, Discrimination and Harassment, Racism, Sexism, Ableism, Religious Hatred, Homophobia, Classism, Body Shaming Threat and Violence.
How we built it
This app was built using Next.js, a popular framework for React. It also leverages the Twitter API and expert.ai NLP API.
Challenges we ran into
I wanted to use the expert.ai in new ways that I haven't seen before. It took some time to formulate this idea and identify the scope of the project. Both APIs used in this project were relatively easy to deal with once I began coding.
Accomplishments that we're proud of
I believe this concept is very useful and the app is easy to use. The app represents a solid proof-of-concept that already inspires further exploration of how we can use NLP tools to help moderate content.
What we learned
I learned a lot about how hate speech detection works specifically. I also came to learn different viewpoints on how we should approach content moderation. I think the expert.ai API does a pretty solid job in identifying true hate speech versus false positives or thoughts/ideas that may be controversial, but do not cross the line into threatening statements.
What's next for Misbehavin'
I'd like to make this more accessible as an API or a tool that can be queried from other apps. It would also be interesting to adapt this idea to work on different social media platforms such as Instagram and TikTok.
Built With
- nextjs


Log in or sign up for Devpost to join the conversation.