Using Word2vec for Music Recommendations – Towards Data Science

Streaming services have changed the way in which we experience content. While recommendation systems previously focused on presenting you with content you might want to purchase for later consumption, modern streaming platforms have to focus instead on recommending content you can, and will want to, enjoy in the moment. Since any piece of content is immediately accessible, the streaming model enables new methods of discovery in the form of personalized radios or recommendation playlists, in which the focus is now more on generating sequences of similar songs that go well together. With now over 700 million songs streamed every month, Anghami is the leading music streaming platform in the MENA region. What this also means, is that the amount of data generated by all those streams proves to be an invaluable training set that we can use to teach machine learning models to better understand user tastes, and improve our music recommendations.