Mining user generated data for music information retrieval

Music is an omnipresent topic in our society as almost everyone enjoys listening to it and many even create it. This is also underlined by the millions of users accessing online social media platforms and services to consume music, among other types of multimedia items. The recent boom of such social media services and the resulting tremendous increase of user generated content (UGC) yielded an enormous amount of this kind of data, which is frequently available through APIs. Even though this data represents a rich source for manifold data mining and information extraction and retrieval tasks, dealing with its noisiness is by no means trivial.