"Minimal defence": a refinement of the preferred semantics for argumentation frameworks
Résumé
Dung's abstract framework for argumentation enables a study of the interactions between arguments based solely on an ``attack'' binary relation on the set of arguments. Various ways to solve conflicts between contradictory pieces of information have been proposed in the context of argumentation, nonmonotonic reasoning or logic programming, and can be captured by appropriate semantics within Dung's framework. A common feature of these semantics is that one can always maximize in some sense the set of acceptable arguments. We propose in this paper to extend Dung's framework in order to allow for the representation of what we call ``restricted'' arguments: these arguments should only be used if absolutely necessary, that is, in order to support other arguments that would otherwise be defeated. We modify Dung's preferred semantics accordingly: a set of arguments becomes acceptable only if it contains a minimum of restricted arguments, for a maximum of unrestricted arguments.
