DSC-Sentinel

Inspiration

We’re big fans of high chat engagement and natural language processing (NLP). Our inspiration for this project came from the idea of merging these interests to analyze chat activity. By following chat engagement in real-time, we can help streamers identify the biggest moments and dull spots, making it easier to capture and highlight the energy of their streams.

What it does

Our Twitch extension monitors the chat for sentiment and engagement, providing streamers with a real-time visual gauge of how their audience is reacting. By analyzing chat messages, we calculate an engagement score between 0 and 1, making it easy for streamers to understand audience sentiment at any moment.

How we built it

The extension’s frontend is built with React and connects to our backend, which is powered by FastAPI. We use Socket.IO for real-time communication between the frontend and backend, allowing for instant engagement updates. The backend is also connected to Twitch's IRC to receive and process live chat messages, making sentiment analysis and engagement tracking possible.

Challenges we ran into

Our biggest challenge was mapping sentiment analysis to accurately reflect chat engagement. Defining what "engagement" means in the context of Twitch chat, and then translating that into meaningful metrics, was a complex task that required careful consideration and fine-tuning.

Accomplishments that we're proud of

We’re proud of building a functional, real-time tool that provides meaningful insights into audience behavior. We successfully implemented a system that not only calculates engagement but does so in a way that feels relevant and useful for streamers.

What we learned

Throughout the project, we gained valuable insights into viewer behavior and what engagement really means in the context of a live chat. We also deepened our understanding of sentiment analysis and its applications within real-time communication systems.

What's next for DSC-Sentinel

Our next steps include refining our algorithm to better capture moments of peak entertainment. We also plan to add features that automatically create clips during high engagement moments or mark these points for streamers, helping them stay connected with their audience and easily capture highlights from their streams.

Built With

• fastapi
• javascript
• python
• react
• sentimentanalysis
• socketsio
• transformers