Fuzzy Unification and Argumentation for Well-Founded Semantics

Argumentation as metaphor for logic programming semantics is a sound basis to define negotiating agents. If such agents operate in an open system, they have to be able to negotiate and argue efficiently in a goal-directed fashion and they have to deal with uncertain and vague knowledge. In this paper, we define an argumentation framework with fuzzy unification and reasoning for the well-founded semantics to handle uncertainty. In particular, we address three main problems: how to define a goal-directed top-down proof procedure for justified arguments, which is important for agents which have to respond in realtime; how to provide expressive knowledge representation including default and explicit negation and uncertainty, which is among others part of agent communication languages such as FIPA or KQML; how to deal with reasoning in open agent systems, where agents should be able to reason despite misunderstandings. To deal with these problems, we introduce a basic argumentation framework and extend it to cope with fuzzy reasoning and fuzzy unification. For the latter case, we develop a corresponding sound and complete top-down proof procedure.