Exploiting User and Item Attributes for Sequential Recommendation

Ke Sun, Tieyun Qian

Abstract: This paper exploits both the user and item attribute information for sequential recommendation. Attribute information has been explored in a number of traditional recommendation systems and proved to be effective to enhance the recommend performance. However, existing sequential recommendation methods model latent sequence patterns only and neglect the attribute information. In this paper, we propose a novel deep neural framework which exploits the item and user attribute information in addition to the sequential effects. Our method has two key properties. The first one is to integrate the item attributes into the sequential modeling of purchased items. The second one is to combine the user attributes with his/her preference representation. We conduct extensive experiments on a widely used real-world dataset. Results prove that our model outperforms the state-of-the-art sequential recommendation approaches.