Setting the Stage: Enhancing Music Recommendations with Spoken Introductions to Stimulate Serendipity
cs.HC
Submitted: 2026-11-03
Updated: 2026-11-03
Code: https://github.com/imec-smit-vub-mep-rec/Music-Introductions-Experimental-Environment
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
The gist: People like to keep their music listening fresh and interesting, but recommender systems often struggle to support this need.
Terminology
Abstract
People like to keep their music listening fresh and interesting, but recommender systems often struggle to support this need. Even when a recommended song is potentially refreshing and enriching, listeners may not perceive it as such. Building on Silvia's appraisal theory of interest, this paper studies whether enhancing recommendations with spoken introductions, immersive or informative, can stimulate two distinct flavors of experienced serendipity: taste-broadening (sparking interest in music outside one's typical preferences) and taste-deepening (deepening interest in music within one's typical preferences). Results from a large-scale user study show that both introduction types stimulate taste-broadening and taste-deepening serendipity. Taste-broadening is most effectively stimulated through transportation (absorption in a narrative world), whereas taste-deepening is most effectively stimulated through cognitive elaboration (learning something new about the artist or the social context in which the music emerged). We discuss how these findings could be translated into practice through curated taste-broadening playlists, where both introduction types appear promising, and taste-deepening playlists, where informative introductions seem particularly effective.
Sources
- Towards Effective Exploration/Exploitation in Sequential Music Recommendation
- Let Me Introduce You: Stimulating Taste-Broadening Serendipity Through Song Introductions
Related papers
- EmoPhone: A Multi-Wave Dataset for In-the-Wild Mobile and Wearable Affect Sensing
- Tasteply: Externalizing Stable and Situational Preference as Manipulable Spatial Objects
- Shielded Trajectory Sampling for Personalized Relational Safety in Conversational AI
- Reconstruction and Reflection of Positive Experiences through Resurfacing Laughter-indexed Everyday Moments
- LLM-Assisted Reranking to Operationalize Nuanced Objectives in Recommender Systems
- Explainable Iterative Data Visualization Refinement via an LLM Agent