Attribute-based Neural Collaborative Filtering. (15th December 2021)

• Record Type: Journal Article
• Authors: Chen, Hai; Qian, Fulan; Chen, Jie; Zhao, Shu; Zhang, Yanping
• Is Part Of: Expert systems with applications. Volume 185(2021)
• Publication Date: 2021-12-15
• Subjects: Recommendation system -- High-order nonlinear -- Attribute information -- Data sparsity; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2021.115539
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: The core task of recommendation systems is to capture user preferences for items. Dot product operations are usually used to mine user preferences for items. However, the dot product can only capture the low-order linear relationships between users and items. In addition, to alleviate the data sparsity problem, current methods mainly introduce auxiliary information, such as user/item attribute information. This attribute information is often treated equivalently. In fact, the importance of this information has different effects on the recommendation results. Therefore, in this paper, we propose a novel Attribute-based Neural Collaborative Filtering (ANCF) method to solve the above problems. Specifically, we use the attention mechanism to distinguish the importance of attribute information and integrate it into the corresponding user and item feature representations to obtain a complete feature representation of users and items. To further capture the high-order interactive relationship between users and items, we use a multi-layer perceptron in ANCF to fully learn the high-order nonlinear relationship between users and items. Extensive experiments on four publicly available datasets demonstrate the effectiveness of the proposed ANCF framework. Highlights: A novel attribute-based neural collaborative filtering method. Use attention mechanism to effectively distinguish user/item attribute information. Explore the linear and non-linear relationship between users and items.