Trust-Object Identification: Approach Towards Trustworthy Objects Identification in S-IoT

The primary goal of Social-Internet-of-Things (SIoT) is to enhance the capabilities of IoT by introducing a social layer, where objects can engage in meaningful relationships, share experiences, and collectively contribute to a more interconnected and intelligent environment. Within the realm of S-IoT, objects possess the capability to form social connections and engage in interactions with other objects. A critical concern demanding attention is the establishment of these relationships in a dependable and trustworthy manner. To address this, a trust framework named Object Identification, based on artificial neural networks, has been conceptualized. Its purpose is to discern the intricate nonlinear connections between input and output for the classification of reliable objects. Object identification is designed to incorporate a range of important trust metrics as input, such as direct trust that combines past and present interactions, the fairness and goodness of an object, the level of relationship, and trustworthy recommendations. This is achieved through the utilization of knowledge graph embedding. Ultimately, comprehensive experiments have been conducted to assess Object Identification’s performance on the dataset. The results show that Object Identification achieves an F1 score of 80%, with lower MAE at 56% and MSE scores at 34%. The results of data analysis depend solely on techniques that extract essential information from the underlying structure of S-IoT with huge uncertainty data embedded in it.