Combining usage and content in an online recommendation system for music in the Long Tail

Abstract

Nowadays, a large number of people consume music from the web. Web sites and online services now typically contain millions of music tracks, which complicates search, retrieval, and discovery of music. Music recommender systems can address these issues by recommending relevant and novel music to a user based on personal musical tastes. In this paper, we propose a hybrid music recommender system, which combines usage and content data. We describe an online evaluation experiment performed in real-time on a commercial web site, specialized in content from the very Long Tail of music content. We compare it against two stand-alone recommender systems, the first system based on usage and the second one based on content data (namely, audio and textual tags). The results show that the proposed hybrid recommender shows advantages with respect to usage-based and content-based systems, namely, higher user absolute acceptance rate, higher user activity rate and higher user loyalty.