Detection of normative conflict that depends on execution order of runtime events in multi-agent systems

Abstract

Norms in multi-agent systems are used as a mechanism to regulate the behavior of autonomous and heterogeneous agents and to maintain the social order of the society of agents. Norms describe actions that must be performed, actions that can be performed and actions that cannot be performed by a given entity in a certain situation. One of the challenges in designing and managing systems governed by norms is that they can conflict with another. Two norms are in conflict when the fulfillment of one causes the violation of the other. When that happens, whatever the agent does or refrains from doing will lead to a social constraint being broken. Several researches have been proposed mechanisms to detect conflicts between norms. However, there is a kind of normative conflict not investigated yet in the design phase, here called runtime conflicts, that can only be detected if we know information about the runtime execution of the system. This paper presents two approaches based on execution scenarios to detect normative conflicts that depends on execution order of runtime events in multi-agent systems. In the first approach, the system designer are able to provide examples of execution scenarios and evaluate the conflicts that may arise if those scenarios would be executed in the system. In the second approach, the conflict checker identifies potential normative conflicts by switching the position order of the runtime events referred in the norm conditions.