Inspired by the 'hedonometer' graph produced by University of Vermont
What it does
Produces a Spotify playlist reflecting the emotions expressed on twitter for a particular region for the past hour. A form of current affairs playback that is purely musical.
How we built it
Python Flask - Machine Learning, we used sentiment analysis to predict the emotions from the tweets from current affairs from different places. We also used multiple APIs like Spotify, Reactiviti AI API to get us the data about the music and the current affairs of the tweets around the world.
Challenges we ran into
APIs, and of how to structure the architecture of the application, and how different APIs with different requirements can evolve your project into something which we have never anticipated of.
Accomplishments that we're proud of
Getting APIs to work, and working in a diverse team and learn a lot about Machine Learning, UI and if something does not go your way, don't fall apart :-)
What we learned
Never to trust APIs (Millie learnt a lot about things other than Java Swing, Gino learned to how to deal with APIs, Harsh learnt how to do sentiment analysis and how to choose the right data, and get more knowledge about the reactiviti API and Pavel learnt how to use the spotify API and how user interfaces can really make a difference)
What's next for Worldify
Possible integration with radio, and more specific region emotion detection, e.g general election emotions etc etc...