Abstract (EN)

Many works have identified the potential benefits of using argumentation in multiagent settings, as a way to implement the capabilities of agents (eg. decision making, communication, negotiation) when confronted with specific multiagent problems. In this paper we take this idea one step further and develop the concept of a fully integrated argumentation-based agent architecture. Under this architecture, an agent is composed of a collection of modules each of which is responsible for a basic capability or reasoning task of the agent. A local argumentation theory in the module gives preferred decision choices for the module’s task in a way that is sensitive to the way the agent is currently situated in its external environment. The inter-module coordination or intra-agent control also relies on a local argumentation theory in each module that defines an internal communication policy between the modules. The paper lays the foundations of this approach, presents an abstract agent architecture and gives the general underlying argumentation machinery minimally required for building such agents, including the important aspects of inter-module coordination via argumentation. It presents the basic properties that we can expect from these agents and illustrates the possibility of this type of agent design with its advantages of high-level of flexibility and expressiveness.