Nonmonotonic Reasoning as Prioritized Argumentation

This paper proposes a formalism for nonmonotonic reasoning based on prioritized argumentation. We argue that nonmonotonic reasoning in general can be viewed as selecting monotonic inferences by a simple notion of priority among inference rules. More importantly, these types of constrained inferences can be speciied in a knowledge representation language where a theory consists of a collection of rules of rst order formulas and a priority among these rules. We recast default reasoning as a form of prioritized argumentation, and illustrate how the parameterized formulation of priority may be used to allow various extensions and modiications to default reasoning. We also show that it is possible, but more diicult, to express prioritized argumentation by default logic: even some particular forms of prioritized argumentation cannot be represented modularly by defaults under the same language.