Computing Argument Preferences and Explanations in Abstract Argumentation

Abstract

We present a novel automated approach for the computation and verification of preferences in an abstract argumentation system. Various argumentation semantics have been developed for identifying acceptable sets of arguments, however, there is a lack of explanatory justification for their acceptability based on preferences. We present an algorithm which takes an abstract argumentation framework and a single extension (conflict-free set of arguments) as input, and outputs preference relations that explain why a set of arguments are acceptable as opposed to their attackers. We also present an algorithm to verify that the output preferences when used with the given argumentation framework induce the input extension.