From User Context to Tailored Playlists: A User Centered Approach to Improve Music Recommendation System.

Systems for recommending music have become a popular tool for delivering track suggestions to users that align with their unique listening habits or preferences. Despite their ubiquity, a recurring focus in these systems has been the accuracy of predictions, often bypassing the impact of user experience (UX) while generating recommendations. The so-called cold-start problem, where the system encounters new users without sufficient data about them, has remained a persistent challenge. This research introduces an approach based on user experience and other aspects that may influence music recommendation (e.g., the cold-start problem, feedback, cognitive effort). The approach, designed for integration into a mobile app, emphasizes context by pre-analyzing user-created playlists for their current context. Our investigation employed the Intermediate Semiotic Inspection Method (ISIM) to evaluate the system’s communicability. This method allowed us to spotlight three categories essential to music recommendation systems: innovative suggestions, constant updates, and users’ engagement in rating. We also apply the Technology Acceptance Model (TAM) to evaluate the system’s acceptance. Our work showcases how a semiotic inspection approach, combined with a focus on user context and feedback, can drastically enhance user experience in music recommendation systems.