Music Recommendation Systems: Techniques, Use Cases, and Challenges

Markus Schedl, Peter Knees, Brian McFee, Dmitry Bogdanov

Abstract: This chapter gives an introduction to music recommender systems, considering the unique characteristics of the music domain. We take a user-centric perspective, by organizing our discussion with respect to current use cases and challenges. More precisely, we categorize music recommendation tasks into three major types of use cases: basic music recommendation, lean-in exploration, and lean-back listening. Subsequently, we explain the main categories of music recommender systems from a technical perspective, including content-based filtering, sequential recommendation, and recent psychology-inspired approaches. To round off the chapter, we provide a discussion of challenges faced in music recommendation research and practice, and of approaches that address these challenges. Topics we address here include creating multi-faceted recommendation lists, considering intrinsic user characteristics, making fair recommendations, explaining recommendations, evaluation, dealing with missing and negative feedback, designing user interfaces, and providing open tools and data sources.