Our own story wasting time reading YouTube comments to decide whether to watch the video or not , because YouTube recently removed the dislikes count feature.
What it does
Analyzes comments on social media and provides a report and a classification of that subject into positive or negative or neutral
How we built it
We made a web browser extension, and a website version using react, the model predicting the sentiment (we used Vader sentiment model because it is optimized for social media texts) on comments is built in a flask server
Challenges we ran into
A lot of comments, heavy computations, many languages, every one writes in his own way without any normalization
Accomplishments that we're proud of
The extension it self, we demonstrated a use case of it in YouTube videos. And also the fact we thought about parallelization of processing on the server.
What we learned
Learned about how vast is this domain is (sentiment analysis) and also about how to make web browser extension
What's next for Comments Analyzer
Right now, it only supports English language and on YouTube, we are willing to generalize it to work in different languages and in different platforms