Recommender systems benefit people at every moment in their daily life. Considerable attentions have been drawn by performance in one-step recommendation and static user-item network, while the performances of recommenders on temporally evolving systems remain unclear. To address this .issue, this p