I've joined twitch recently, for the most part I see streamers being friendly and can inspire future generations, but occasionally I've come across a lot of negative streamers as well, I don't think that provides a positive example for their viewers or the community, especially the younger generation.
What it does
By using NLP Sentiment Analysis on all the streamer's captions, we get a closer view on whether the streamer is giving positive, neutral or negative influence to their viewers. All of the captions are being analyzed, scored and displayed into a pie chart which updates in real time as new captions are coming out. This component gives 2 major advantages, when viewer sees the stream is being too negative they can op out, when streamer can also react when they see their own analysis.
How I built it
I used node.js as backend, which hosts on amazon severless, this does the sentiment analysis hits in real time. We built a stream component with a d3 charts to display the scores when the results come back.
Challenges I ran into
Building and integrating twitch was not easy.
Accomplishments that I'm proud of
After a few days I was able to get twitch component running and bulit out my first twitch extension that I'd like to use.
What I learned
When people realize the sentiment is negative, they will walk away.
What's next for Streamer Sentiment Analysis
It's publishing in the store and we want to keep improving the langauge