The outcomes of logic-based argumentation systems under preferred semantics - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2012

The outcomes of logic-based argumentation systems under preferred semantics

Résumé

Logic-based argumentation systems are developed for reasoning with inconsistent information. They consist of a set of arguments, attacks among them and a semantics for the evaluation of arguments. Preferred semantics is favored in the literature since it ensures the existence of extensions (i.e., acceptable sets of arguments), and it guarantees a kind of maximality, accepting thus arguments whenever possible. This paper proposes the first study on the outcomes under preferred semantics of logic-based argumentation systems that satisfy basic rationality postulates. It focuses on systems that are grounded on Tarskian logics, and delimits the number of preferred extensions they may have. It also characterizes both their extensions and their sets of conclusions that are drawn from knowledge bases. The results are disappointing since they show that in the best case, the preferred extensions of a system are computed from the maximal consistent subbases of the knowledge base under study. In this case, the system is coherent, that is preferred extensions are stable ones. Moreover, we show that both semantics are useless in thic case since they ensure exactly the same result as naive semantics. Apart from this case, the outcomes of argumentation systems are counter-intuitive.
Fichier principal
Vignette du fichier
ID50.pdf (243.82 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04305191 , version 1 (24-11-2023)

Identifiants

Citer

Leila Amgoud. The outcomes of logic-based argumentation systems under preferred semantics. 6th International Conference on Scalable Uncertainty Management (SUM 2012), Sep 2012, Marburg, Germany. pp.72--84, ⟨10.1007/978-3-642-33362-0_6⟩. ⟨hal-04305191⟩
32 Consultations
10 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More