Using default logic for agent behavior testing

An agent is an autonomous entity that can perform actions to achieve its goals. It acts in a dynamic environment that may engender failures regarding its behavior. Therefore, a formal testing/verification approach of the agent is required to ensure the correctness of its behavior. In this paper, we propose a Default Logic formalism to abstract an agent behavior as knowledge and reasoning rules, and to verify and test the consistency of the behavior. The considered agents are implemented with JADE framework. Also, agent abstraction is translated into Answer Set Programming and solved by Clingo to generate dynamic and adaptive test cases of the agent behavior. The dynamic test cases allow predicting the agent behavior when a new information arises in the system.