Web Dev / Personal Project
Twitch Insights
Follow the mood of a live Twitch chat through sentiment timelines, visual breakdowns, and exportable reports.
How it works
The application captures live Twitch chat through the Twitch API. A RoBERTa model on a Flask backend classifies messages as positive, negative, or neutral. MongoDB stores the results, and a React dashboard displays sentiment timelines, mood distributions, and word clouds for streamers and analysts exploring audience engagement.
Capabilities
- Live chat sentiment classification with RoBERTa
- Sentiment timelines and mood distribution charts
- Word clouds and exportable reports
- Twitch API chat integration
- Sentiment data stored in MongoDB
Built with
- React
- Flask
- MongoDB
- Python