Planning with argumentation schemes in online dispute resolution

Online dispute resolution is becoming the main method when dealing with a conflict in e-commerce. Our interest regards automating the dispute resolution process in order to face the increasing number of disputes. Quite aware of the difficulty that lies ahead of such task, we drive our attention to argumentation schemes to cover the gap between arguments based on propositional logic and plain linguistic arguments used by the human mediators. The contribution here consists in proposing a classification of the argumentation schemes suitable for B2B disputes. Finding the argumentation line supporting a claim is considered as an AI planning problem and it is meant to be agent-driven. By using argumentation schemes we intend to maintain a high level of abstraction to easily accomodate human intervention. The schemes are formalising in PDDL in a framework compatible with the emerging semantic web. The main advantages relie on the facts that communication is guided by the critical questions, whilst PDDL offers metrics to compare argument chains under time constraints.