Managing Cold-Start Issues in Music Recommendation Systems: An Approach Based on User Experience

Abstract

Music recommendation systems have been widely used to suggest songs to users based on their listening history or interests. Traditionally, most recommender systems have focused on prediction accuracy without considering user experience (UX) in generating recommendations. In addition, there is also the problem of cold-start, which is when the system has new users and not enough data is available about them. This study presents a new approach for music recommendation based on user experience that explores the cold-start problem. We implemented our approach in a mobile application and evaluated the system’s communicability using the Intermediate Semiotic Inspection Method (ISIM). As a result, we identified three categories relevant to music recommendation systems: novelty in recommendations, continuous updates, and users’ interest in rating. In addition, we checked each participant’s understanding of the tool, which was generally very close to the intended proposal.