What it does
Detects toxic language and filters it out in Discord Servers using natural language processing from Cohere
How we built it
We used Cohere's NLP for toxicity detection and Discord for the bot.
Challenges we ran into
When testing the initial version of our bot, we realized there were several toxic phrases it wasn't detecting. We solved this problem to an extent by training the model with another dataset we found online, which contained tweets with a variety of toxicity values. Even after implementation of the new model, the bot wasn't picking up every toxic message.
Accomplishments that we're proud of
We're proud that we were able to use Cohere's NLP to create something that makes online spaces safer.
What we learned
We learned that no model will work perfectly. Even with hundreds of thousands of data entries, a model will not be able to perfectly identify every instance of toxic language.
Built With
- cohere
- discord
- machine-learning
- python
- streamlit
Log in or sign up for Devpost to join the conversation.