A bio-inspired trust prediction approach in time series of varying characteristics

With the growth of technology and increasing interest in communicating widely, the concept of trust become more significant. Trust in a mutual relationship refers to the subjective belief of one agent about another. In each experiment, the trusting agent assigns a value to trustee agent which represents the satisfaction level of the agent. The time between two experiments is called the time slot, and predicting the trust value of future time sluts helps partners to identify the probability of continuing their relationship in future. To predict these values, it is necessary to apply historical data. This work addresses a unique simulation structure of historical trust values between two partners which considers all possible permutations of the trust modes. A mutual relation which has lasted for 21 months is considered, based on which 108 different scenarios of trust modes are proposed. The modes are categorized as high trust, medium trust and low trust. To predict the trust values of next slot, two methods, Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN), are implemented. By comparing both methods, the appropriate prediction tool is identified.