Learning from History and Present

In the modern e-commerce, the behaviors of customers contain rich, e. g. , consumption habits, the dynamics of preferences. Recently, -based recommendations are becoming popular to explore the temporal of customers' interactive behaviors. However, existing works exploit the short-term behaviors without fully taking the customers'-term stable preferences and evolutions into account. In this paper, we a novel Behavior-Intensive Neural Network (BINN) for next-item by incorporating both users' historical stable preferences and consumption motivations. Specifically, BINN contains two main, i. e. , Neural Item Embedding, and Discriminative Behaviors Learning. , a novel item embedding method based on user interactions is developed obtaining an unified representation for each item. Then, with the embedded and the interactive behaviors over item sequences, BINN discriminatively the historical preferences and present motivations of the target users. , BINN could better perform recommendations of the next items for the users. Finally, for evaluating the performances of BINN, we conduct experiments on two real-world datasets, i. e. , Tianchi and JD. The results clearly demonstrate the effectiveness of BINN compared several state-of-the-art methods.