Show Notes
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 Rebecca's work is the Music Semantics dataset, created by scraping Reddit discussions to capture how people naturally describe music using atmospheric qualities, contextual comparisons, and situational associations rather than just technical features. This dataset, available on Hugging Face, enables more nuanced recommendation systems that better understand user preferences and support niche tastes. Her research utilizes industry datasets including Last.fm and Spotify's Million Playlist Dataset, and points toward exciting future applications in music generation and multimodal systems that combine audio, text, and video.
More from Data Skeptic
The next evolution of wearable technology may not require new screens or controllers, but new ways to interact with the devices we already wear. Shimon explores how gesture-sensing earbuds could transform virtual reality and wearable computing through intuitive, user-centered interactions that make technology more accessible and adaptable.
How do we design robots and autonomous vehicles that understand the unwritten rules of human behavior? Kyle speaks with Cornell Tech professor Wendy Ju about implicit interaction, "Wizard of Oz" prototyping, and what studying pedestrians, self-driving cars, and even robotic furniture can teach us about designing technology that behaves the way people expect.
The data we collect about ourselves can tell us a lot—but only if the technology collecting it actually fits into our lives. Daniel Epstein explores personal informatics, from fitness trackers and food journals to baby tracking and AI, and explains why abandoning a tracking tool doesn't necessarily mean it failed.
In the final episode of our Recommender Systems season, we explore the growing questions of trust, manipulation, privacy, fairness, sustainability, and user control. From fake reviews and shilling attacks to explainable recommendations and user-selected algorithms, we look at what happens when recommender systems must answer not only for what they recommend, but for the consequences of those choices.
In part two of the Data Skeptic Recommender Systems season finale, Kyle asks a deceptively difficult question: what should recommender systems actually optimize for? Drawing on conversations from across the season, the episode explores engagement, filter bubbles, popularity bias, fairness, human curation, embeddings, and the growing role—and risks—of large language models in shaping what gets recommended to us.