Ontology-based Recommender System for the Million Song Dataset Challenge

To deal with the Million Song dataset challenge, many studies have been invested in music recommender systems. This dataset is practical and popular in music information retrieval, thus also investigated in this study. This paper proposes an ontology-based music recommender system which can overcome semantic problems in song datasets and improve music recommendation making. Some ontology models are constructed to efficiently organize the Song dataset. And some ontology reasoning strategies are developed to make music recommendations with high accuracy. Some experiments are conducted to evaluate the proposed recommender system and compare with previous baseline experimental results.