Particularly, we also developed an App hash tree to efficiently look up Apps in recommendation. Finally, the experiments on a large-scale real-world data set clearly validated the effectiveness and efficiency of the proposed recommendation framework.