A logical framework for modeling argumentation using labelled deduction

Classical methods for representing and reasoning with knowledge rely on the assumption that the available information is complete, certain and consistent. In real-world problems this is usually not the case, and Al has long dealt with the issue of finding a suitable formalization for commonsense reasoning. Defeasible argumentation [SL92, CMLOO, PV99] has proven to be a successful approach in many respects, since it naturally resembles many aspects of human commonsense reasoning. Our intention is to find a logical framework in which the diverse aspects of defeasible argumentation can be formally captured, in order to analyze their emerging properties. The issue of defining a logical framework for defeasible argumentation with labels has been tackled before in alternative ways. Hunter proposed a framework for characterizing structural information using labelled formulas combined with argumentation [Hun94]. Fox & Parsons [FP97] defined a Logic of Argumentation, a qualitative approach to decision making which makes use of labelled formulac, presented as an altemative to standard formalisms in order to overcome some of the limitations imposed by them. Our approach focuses on formalizing an argunentative framework using Defeasible Logic Programming [Gar97] as a theoretical basis, combined with labelled deductive systems [Gab96]. In this presentation we describe the main aspects of our formalization. Eje: Aspectos teóricos de inteligencia artificial Red de Universidades con Carreras en Informática (RedUNCI)