An axiomatic account of preference-based argumentation systems

Leila Amgoud (1, 2) , Srdjan Vesic (1)

1 IRIT-ADRIA - Argumentation, Décision, Raisonnement, Incertitude et Apprentissage
2 CNRS - Centre National de la Recherche Scientifique

Argumentation is a reasoning model based on the construction and evaluation of arguments. Dung has proposed an abstract argumentation framework in which arguments are assumed to have the same strength. This assumption is unfortunately not realistic. Consequently, three main extensions of the framework have been proposed in the literature. The basic idea is that if an argument is stronger than its attacker, the attack fails. The aim of the paper is twofold : First, it shows that the three extensions of Dung framework may lead to unintended results. Second, it proposes a new approach that takes into account the strengths of arguments, and that ensures sound results. We start by presenting two minimal requirements that any preference-based argumentation framework should satisfy, namely the conflict-freeness of arguments extensions and the generalization of Dung’s framework. Inspired from works on handling inconsistency in knowledge bases, the proposed approach defines a binary relation on the powerset of arguments. The maximal elements of this relation represent the extensions of the new framework.

Mots clés: Argumentation; Preferences

Domaines: Intelligence artificielle [cs.AI]

Leila Amgoud, Srdjan Vesic. An axiomatic account of preference-based argumentation systems. 5èmes Journées francophones sur les Modèles Formels de l'Interaction (MFI 2009), Jun 2009, Lannion, France. ⟨hal-03301296⟩