Abstract:

This paper addresses the cold-start problem based on the context of new users'playing music.The decision tree classification model is constructed with the context information of time，weather，state when users play music.When new users use the music recommendation system，their context information is extracted and their interests are estimated according to the branch of the decision tree and the context information so that certain music is recommended to them.The experimental results show that when the recommended music does not exceed six items，new users are more satisfied with the recommendations than with the recommendations offered by TopN.

Key words: music recommendation, decision tree classification, user cold-start