Potential Relationship Discovery in Tag-Aware Music Style Clustering and Artist Social Networks

With the rapid growth of music information and data in today’s ever changing world, exploring and analyzing music style has become more and more difficult. Traditional content-based methods for music style analysis and newly emerged tag-based methods usually assume music items are independent of each other. However, in real world applications, do there exist some relationships among them. In this paper, we construct the social relation graph among different music artists by extracting the friendship information from social media such as Twitter, and incorporate the generated social networking graph into tag-based music style clustering. Experiments on real data show the effectiveness of this novel integration of different information sources.