On the use of contexts for representing knowledge in defeasible argumentation

The notion of context and its importance in knowledge representation and nonmonotonic reasoning was first discussed in Artificial Intelligence by John McCarthy. Ever since, contexts have found many applications in developing knowledge-based reasoning systems. Defeasible argumentation has gained wide acceptance within the Al community in the last years. Different argument-based frameworks have been proposed. In this respect, MTDR (Simari & Loui, 1992) has come to be one of the most successful. However, even though the formalism is theoretically sound, there exist sorne dialectical considerations involving argument construction and the inference mechanism, which impose a rather procedural approach, tightly interlocked with the system's logic. This paper discusses different uses of contexts for modelling the process of defeasible argumentation. We present an alternative view of MTDR using contexts. Our approach will allow us to discuss novel issues in MTDR, such as defining a set of moves and introducting an arbiter for regulating inference. As a result, protocols for argument generation as well as some technical considerations for speeding up inference will be kept apart from the logical machinery underlying MTDR.Eje: 2do. Workshop sobre aspectos teóricos de la inteligencia artificialRed de Universidades con Carreras en Informática (RedUNCI