Distributional Models vs. Linked Data: leveraging crowdsourcing to personalize music playlists

The document discusses the development of Play.me, a system designed to create personalized music playlists by extracting user preferences from social media, particularly Facebook. It compares two enrichment techniques: content-based strategies and distributional models, which leverage linked data and semantic relationships to enhance playlist suggestions. The authors argue that current recommendation systems face challenges in training and user interest information, and they propose improvements using crowdsourcing and explicit user data.