Combining audio content and social context for semantic music discovery