Want to create an interactive transcript for this episode?
Podcast: Data Skeptic
Episode: Music Playlist Recommendations
Description: In this episode, Rebecca Salganik, a PhD student at the University of Rochester with a background in vocal performance and composition, discusses her research on fairness in music recommendation systems. She explores three key types of fairness—group, individual, and counterfactual—and examines how algorithms create challenges like popularity bias (favoring mainstream content) and multi-interest bias (underserving users with diverse tastes). Rebecca introduces LARP, her multi-stage multimodal framework for playlist continuation that uses contrastive learning to align text and audio representations, learn song relationships, and create playlist-level embeddings to address the cold start problem. A significant contribution of R...