UIAE: Collaborative Filtering for User and Item based on Auto-Encoder
Hang Zheng, Xing Xing, Qiuyang Han, Mindong Xin, Yong Niu · 2021 7th Annual International Conference on Network and Information Systems for Computers (ICNISC) · 2021
These years, deep learning achieves huge success on recommender systems. Collaborative filtering using past interaction information is the primary method for recommender system. Although some recent works employ deep learning for collaborative filtering, they only consider one dimension of interaction information. We present a new model called collaborative filtering for users and items based on Auto-Encoder (UIAE). It learns interaction information from two dimensions. UIAE is a union of User-based and Item-based collaborative filtering that gets better performance than each of them. Experiments on two data sets are made researching the impacts of different components of UIAE and comparison experiments with previous CF methods based on Auto-Encoder are made to prove our model has better performance.