Recommender systems are playing an increasingly important role in e-commerce portals. Based on the massive data from, we are building a novel recommendation model into the one  of largest e-commerce platform with the most advanced technologies in the industry.  Our recommendation model has been applied on the JD mall and JD app to help billions of JD users.

Meizi Zhou, Zhuoye Ding, Jiliang Tang, Dawei Yin, Micro Behaviors: A New Perspective in E-commerce Recommender Systems, In WSDM’18, Los Angeles, USA, 2018.

The explosive popularity of e-commerce sites has reshaped users’ shopping habits and an increasing number of users prefer to spend more time shopping online. This evolution allows e-commerce sites to observe rich data about users. The majority of traditional recommender systems have focused on the macro interactions between users and items, i.e., the purchase history of a customer. However, within each macro interaction between a user and an item, the user actually performs a sequence of micro behaviors, which indicate how the user locates the item, what activities the user conducts on the item (e.g., reading the comments, carting, and ordering) and how long the user stays with the item. Such micro behaviors offer fine-grained and deep understandings about users and provide tremendous opportunities to advance recommender systems in e-commerce. However, exploiting micro behaviors for recommendations is rather limited, which motivates us to investigate e-commerce recommendations from a micro-behavior perspective in this paper. Particularly, we uncover the effects of micro behaviors on recommendations and propose an interpretable Recommendation framework RIB, which models inherently the sequence of m{\bf I}cro {\bf B}ehaviors and their effects. Experimental results on datasets from a real e-commence site demonstrate the effectiveness of the proposed framework and the importance of micro behaviors for recommendations.

Figure 1 illustrates a real example of observed data on a user from an e- commerce site in a short period.

The architecture of our framework is shown in Figure 2. It consists of five layers – an input layer, an embedding layer to solve the sparse and high-dimensional challenge, a RNN layer to model sequential information, an attention layer to capture varied effects of micro behaviors and an output layer.

Figure 2 e architecture of the proposed framework.