---

Personalized Recommendation Engine

Methodology

A personalized music recommendation engine uses a statistical analysis of your music listening tendencies and proclivities to recommend music to the listener. There are a variety of ways to collect information about a listener's likes and dislikes, including analyzing playlists, playcounts, and ratings, as well as genre tendencies and tags made by the listener. A personalized recommendation engine analyzes this information and uses it along with information provided by other users, music reviewers, and/or audio analysis to create music recommendations that the listener is statistically probable to enjoy.

Pros and Cons

A personalized music recommendation engine is the ideal way for listeners to find out about the music they are most likely to appreciate and enjoy. Recommendations are specifically tailored to the listener so that he/she can efficiently discover the music in which he/she is likely to be interested. Services such as Last.fm, LAUNCHcast, and Audiobaba all are different implementations of this general idea. The primary problem faced by these engines is the cyclical nature of their recommendations. For example, if a listener only listens to electronic music, he will be directed to more electronic music, and thus might not discover his hidden passion for jazz music. This cyclical nature also makes it difficult for unknown artists to gain popularity, because, as they stand now, personalized recommendation engines direct the listener to what other similar listeners enjoy, not what they would enjoy but haven't yet discovered.