Abstract Dialectical Frameworks Semantics, Discussion Games, and Variations

Argumentation is an essential  part of our daily life both in our individual and our social activities. The extensive trajectory of research on argumentation from Aristotle to today's computational argumentation in artificial intelligence  shows  how far research in argumentation has come. Argumentation theory can shed light on the process of decision making, from modeling to evaluating a problem. Models of argumentation reflect how arguments  relate to one another, and semantics of models of argumentation reflect how to use argumentation for making a decision under inconsistent, controversial, and incomplete information.

In this thesis we consider abstract dialectical frameworks (ADFs for short), one of the powerful formalisms of argumentation. The semantics  of  ADFs  are  methods  proposed to evaluate the acceptance of the arguments. We begin by focusing on the semantical evaluation of ADFs, presenting two novel semantics, namely the strongly admissible and semi-stable semantics. Next, we introduce the first discussion games for ADFs to provide an explanation of why an argument of interest is to be accepted  according to a given semantics. Furthermore, we present subclasses of ADFs and we investigate how the restrictions that we consider influence the semantic evaluation of such ADFs. Next, we combine argumentation with decision theory in the context of ADFs in order to model expected utility problems. With this work, we hope that we have advanced the knowledge on the field of formal argumentation.

Dissertation: http://hdl.handle.net/11370/e0121ee8-35f6-402c-9421-401614aa4d13