Tuning Evidence-Based Trust Models

Many evidence-based trust models require the adjustment of parameters such as aging- or exploration-factors. What the literature often does not address is the systematic choice of these parameters. In our work, we propose a generic procedure for finding trust model parameters that maximize the expected utility to the trust model user. The procedure is based on game theoretic considerations and uses a genetic algorithm to cope with the vast number of possible attack strategies. To demonstrate the feasibility of the approach, we apply our procedure to a concrete trust model and optimize the parameters of this model.