Abstract

Music recommender systems have become a crucial part of several music players, like Spotify, Pandora, Apple Music, and Google Play Music. Music recommender systems are inherently more complex than other recommender systems (like recommender systems for shopping) because of the subjective nature of music and the ability for users’ preferences to modulate with mood. In this talk, we will define the components of a music recommender system and discuss several approaches to music recommending. We will also explore one of the more popular approaches, a content-based algorithm, in more depth. Finally, we will dive into a case study of Spotify’s recommender system, by taking a look at its automatically curated Discover Weekly playlists that are personalized for each user.