Heterogeneous Multi Relation Trust for SIoT Service Recommendation

Geming Xia, Chaodong Yu, Linxuan Song, Wei Peng, Yuze Zhang, Hongfeng Li

Abstract: With the rapid development of Internet of Things (IoT) technology, the deep integration of IoT devices and social networks has given rise to the field of Social Internet of Things (SIoT), where the significance of SIoT service recommendation is increasingly prominent. Given the heterogeneity and complexity of user interaction patterns in the SIoT service recommendation, as well as the abundance of user-centric semantic information, building an effective trust mechanism has become a key element in improving service quality and promoting successful interaction between users. In this paper, we propose a novel heterogeneous multi relation trust evaluation model (HMTrust). This model accurately captures the association patterns of users in multi-dimensional trust communities based on meta-paths. Subsequently, we introduce a two-layer attention mechanism to delve into the rich semantic features embedded within heterogeneous trust relationships, achieving a refined evaluation of trust relationships. Based on this evaluation result, we can effectively implement personalized SIoT service recommendation to meet the diverse needs of users. Experiments on real-world datasets show the advantages of our proposed HMTrust compared to existing baselines.