An approach for detecting deception in agents

The role of intelligent agents has been recognized as important in various competitive and complex application environments. Since the intelligent agent is developed to be cleverer, more complex, and yet uncontrollable, a number of problems have recently been recognized. Examples of theses problems are: how to identify whom we can trust, how we know which agent is lying and how to manage information that is acquired from them. We propose two related techniques to deal with some of theses issues. The first relates to the trust measurement of agents and the second involves detecting a deception approach that is able to adjust weights by group and by individual. With regard to evaluation of proposed techniques, the first experiment has been conducted based on comparing the results generated by using existing trust models and the proposed technique, where the result clearly shows that the proposed technique provides a better, more accurate way to detect and evaluate an agent.