In this paper, we present a Web recommender system for
recommending, predicting and personalizing music playlists based on a user
model. We have developed a hybrid similarity matching method that combines
collaborative filtering with ontology-based semantic distance measurements.
We dynamically generate a personalized music playlist, from a selection of
recommended playlists, which comprises the most relevant tracks to the user.
Our Web recommender system features three functionalities: (1) predict the
likability of a user towards a specific music playlist, (2) recommend a set of
music playlists, and (3) compose a new personalized music playlist. Our
experimental results will show the efficacy of our hybrid similarity matching
approach and the information personalization method