Musical data mining for electronic music distribution

Abstract

Music classification is a key ingredient for electronic music distribution. Because of the lack of standards in music classification (or the lack of enforcement of existing standards), there is a huge amount of unclassified music titles in the world. The authors propose a classification method based on a musical data mining technique based on co-occurrence and correlation analysis that can be used for classification. It gives a new approach to similarity between several titles of music or several artists. We study large corpora of textual information referring titles of music or artists whose names are decided by humans without particular constraints other than readability, and draw various hypotheses concerning the natural similarities that emerge from these corpora. Based on a clustering technique, we show that interesting groups can reveal specific music genres and allow classification of music titles in an objective manner.