Inspiration
With social media being more polarizable than it's ever been, the rise of hate speech on the internet is alarming. This is especially detrimental for online communities, whose organizers don't have a proper way to manage and control hate speech on their communication channels.
Until now.....
What it does
NoHateBot is a discord bot that identifies identity-hate speech in a server. Once identity-hate speech is done by a user, our custom model classifies it as either hate or not. If it is an identity hate speech our bot flags the user, deletes the message, and sends the server's owner a direct message on discord regarding the incident. All of this is performed in real-time with minimal Delay. There are also 2 types of classification models we are using: one is a custom model we built and another is co:here platform's classify classification model.
How we built it
We built the project using python. The discord bot is made using discord.py which is an API wrapper for Discord. There are 2 models used. One is a custom model built using scikit-learn and nltk. The other is co:here platform's integration model. The website for the discord bot is built using html5, css3, javascript and bootstrap.
Challenges we ran into
- Increasing the accuracy of our custom model and integrating it with real-time discord messages was one of the major challenges, we ran into. To remedy this, we are currently training the model every time bot is restarted. Once the model is trained and ready, incoming discord messages can be classified in real-time.
Accomplishments that we're proud of
We learned more about co:here platform and Circle CI. We also learned how to integrate NLP models with discord.
What's next for NoHate Bot
- More ML models promoting inclusiveness in a server
- Better classification Strategies.
Built With
- bootstrap
- cohere
- css3
- discord.py
- html5
- javascript
- nltk
- python
- scikit-learn

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