Blooming Beats: An interactive explainable recommender system for exploring personal music narratives through data humanism principles

Abstract

Recommender systems have become integral to our daily digital experiences, particularly in music streaming services. However, existing systems often reduce rich listening experiences to mere analytics, overlooking the personal narratives and contexts that make music meaningful. We present Blooming Beats, an interactive explainable recommender system that transforms a decade of personal Spotify listening data into compelling visual narratives. Grounded in Data Humanism principles, it represents songs as flower-like graphs where petals encode audio features, while connecting lines and contextual markers capture listening behaviors, personal milestones, and global events. The system enables users to explore 10 years of listening history through multiple temporal views, re-live musical moments, and generate context-aware recommendations. Through a preliminary qualitative user study, we found that Blooming Beats enables users to explore listening histories and understand recommendations in tailored and contextualized ways, demonstrating how Data Humanism can be employed to engage with personal data to enhance music recommendation explainability.