An MPD Player with Expert Knowledge-basedSingle User Music Recommendation

Hen-Hsen Huang

Abstract: This work demonstrates a music player based on Music Player Daemon (MPD), a protocol popular for audiophiles, with personalized music recommendation. As a standalone player, we focus on the scenario of single user recommendation. Different from collaborative filtering based recommendation, which relies on usage patterns from a large number of users, we propose a novel approach that does not require other users’ information. We formulate the recommendation as a task of knowledge base completion and exploit the expert knowledge from a music knowledge base. The effectiveness of our approach is evaluated, and the player is released as an open-source software for music lovers.