A Note on Content-based Collaborative Filtering of Music

Collaborative filters are frequently used in e-commerce to provide a heightened user experience and to tempt users into making purchases by recommending items and drawing the user’s attention to additional products. Purchasing of digital media over the Internet continues to be popular and e-commerce giants such as Amazon.com, CDNOW.com and Launch.com heavily employ Automated Collaborative Filtering (ACF). This paper demonstrates a system for comparing musical compositions and provides an indication of how similar two or more musical pieces are to each other. It is shown that a significant amount of similarity exists between music compositions analysed from within the same genre. It is proposed that a similarity metric could be incorporated into existing ACF systems to provide a powerful and effective recommendation system that will cater specifically for a user’s preferences, and thus encourage purchase.