Policing Virtual Organizations

Autonomous agents within a multi-agent system can undertake behaviors that, while profitable for them, are damaging to the system as a whole. One way of regulating an agent’s behavior is by obliging it to act in accordance with a set of specifications set out in a contract, and sanctioning it otherwise. In this paper, We propose a framework that performs two major subtasks of contract based behavior enforcement in partially observable multi-agent environments: behavior specification and behavior violation detection. The framework consists of three main components. The first is a contract specification language used to specify desired agent behavior. The second is a dialogue game used to present and refute evidence about an agent’s actual behavior. The final component of the framework is a mechanism used to combine the evidence so as to reach a decision regarding an agent’s guilt or innocence. We also examine existing approaches for performing contract enforcement, and describe a concrete environment where investigations into the effects of policing can take place.