In this paper, we consider a multi-agent system where agents perform argumentation activity on the basis of knowledge both stated in their knowledge-bases and acquired from other agents. Starting from a previous proposal, we introduce a distributed approach to argumentation, removing some limiting assumptions, and we show by means of an example that the system can not be guaranteed to reach a state which satisfy a well-founded semantics, and not even to terminate. In order to tackle this problem, we devise two self-stabilizing algorithms for the computation of the defeat status, one specifically tailored to rebutting defeat, and the other able to deal with any form of defeat.
PDF file