Collaborating Aesthetic Change and Heterogeneous Information into Recommender Systems

Recently, with the increasing of heterogeneous information, recommender system has gradually transferred from a single view of rating to multi-dimensional information integration. However, the existing approaches cannot fully exploit the users’ information. In this paper, we propose a deep learning based framework, which uses heterogeneous information and also considers temporal changes in users’ interests to extract the users’ features. Experiments on several real world datasets demonstrate that our proposed model outperforms the state-of-the-art methods for recommendation task.