Formalizing Multi-Agent Systems Using Action Descriptions in Single Agent Perspective

Abstract

Logic-based representations of multi-agent systems have been extensively studied. In this work, we focus on the action language BC to formalize global views of MAS domains. Methodologically, we start representing the behaviour of each agent by an action description from a single agent perspective. Then, it goes through two stages that guide the modeler in composing the global view by first designating multi-agent aspects of the domain via potential conflicts and later resolving these conflicts according to the expected behaviour of the overall system. Considering that representing single agent descriptions is relatively simpler than representing multi-agent description directly, the formalization developed here is valuable from a knowledge representation perspective.