Music Recommendation with Collaborative Filtering for Mobile Services

As the development of the mobile communication and the computational capability of the mobile terminals, more users use their mobile devices to play music. In this work, an online music recommendation system is designed for mobile services, which consists of two modules: offline processing and online recommendation. The offline module labels all the music into different categories, by which the music items libraries corresponding to the tags are constructed and the rating matrixs are consequently built. The online module integrates the context information, by which the matched rating matrix is retrieved. By using the collaborative filtering model with matrix completion algorithm, the music recommendations that suit the user and the situation are offered. The proposed recommendation system improves the precision of the recommendation by integration the context information of the users, and augments the online computational capability because the matrix scale is reduced by constructing the rating matrices for the music in the different tag libraries. A large number of experiments demonstrate that the proposed system is designed to be robust and effective to the music recommendation and efficient to the online recommendation for the mobile services.