A Privacy-Preserving Cross-Domain Recommendation Algorithm for Industrial IoT Devices
Xu Yu, Qinglong Peng, Hongwu Lv, Dingjia Zhan, Qiang Hu, Junwei Du, Dunwei Gong
Abstract
Recommendation algorithms have been initially applied on the online business platform of industrial Internet of Things (IoT) devices. However, traditional recommendation algorithms are often difficult to solve the data sparsity problem. In fact, online shoppers are often accompanied by consumption behavior of other heterogeneous products, so we combine the consumer behavior of other heterogeneous products in the auxiliary domain to improve the recommendation performance of industrial IoT devices in the target domain. Due to privacy-preserving requirements, the original scoring information of the auxiliary domain is often not allowed to be directly shared with the target domain. Therefore, we propose a Privacy-Preserving Cross-Domain Recommendation algorithm for industrial IoT devices. First, the non-privacy preference features are extracted through the auxiliary domain scoring data. Next, the extracted preference features are fused with the target domain information. Extensive experiments have been conducted on the Amazon dataset to verify the effectiveness of our method.