Can Argumentation Help AI to Understand Explanation?

ed from context by seeing an explanation inferentially, much in the same way that early expert systems saw an explanation as chaining of inferences. This approach omitted, for the most part, to see the purpose of an explanation as increasing understanding, because (a) it was directed to the analysis of explanation in the natural sciences rather than to the study of explanation in everyday reasoning, and (b) because it saw the notion of understanding as not amenable to precise logical/scientific analysis. Recent work in cognitive science, on the other hand, has postulated that the aim of an explanation is to increase understanding, and argued that explanations fail when they do not increase the understanding of the phenomenon the purport to explain. Recent research in argumentation and artificial intelligence (Dunn and Bench-Capon, 2006) is now based on such dialog structures in which two parties reason together to work towards attaining a communicative goal. Thus it is natural to look ...

• Type: Journal
• Year: 2008
• Where: KI
• Authors: Doug Walton
• Keywords: Artificial Intelligence, Everyday Reasoning, Explanation, KI 2008, Model