Designing (more) Personalized Music Recommendation Systems: An Architecture for Integrating Interaction Context and User Experience

Music recommender systems are pivotal to streaming platforms, yet personalization typically relies on listening histories, overlooking context, and user experience. This gap motivates our research objective: to design an architecture that integrates contextual inputs, such as user activity, and explicit UX feedback into the recommendation process. We introduce UConteXt Arch, an architecture layered with Interface, Middleware, Service, and Data tiers that capture context and feedback, and adapt recommendation models. We implemented UConteXt Arch in MixFy, a proof-of-concept Flutter application integrated with Spotify’s Web API to demonstrate feasibility. We conducted two evaluations: a semiotic inspection with eight participants to verify interpretation of context-aware recommendations and refine interface elements. A five-day longitudinal study with 44 participants using the UX Curve and the Technology Acceptance Model measures to assess acceptance and impact. Results show clear user understanding of context adaptations, significant increases in perceived usefulness, ease of use, and intention to continue use, alongside music discovery aligned with user activities. These findings confirm that embedding contextual sensing and UX-driven feedback within a unified architecture enhances recommendation relevance and user engagement, providing a foundation for future research on context-aware recommender systems.