Empirical Argumentation: Integrating Induction and Argumentation in MAS

This paper presents an approach that integrates notions and techniques from two distinct fields of study —namely inductive learning and argumentation in multiagent systems (MAS). We will first discuss inductive learning and the role argumentation plays in multiagent inductive learning. Then we focus on how inductive learning can be used to realize argumentation in MAS based on empirical grounds. We present a MAS framework for empirical argumentation, A-MAIL, and then we show how this is applied to a particular task where two agents argue in order to reach agreement on a particular topic. Finally, an experimental evaluation of the approach is presented evaluating the quality of the agreements achieved by the empirical argumentation process.