Extending Defeasible Logic Programming with Informant-based Argumentation

Abstract

This work introduces an informant-based argumentative approach for deliberative agents that can receive domain knowledge from other agents in a multi-agent setting. Also, agents will be able to model, through the use of backing and detracting rules, reasons for and against the consideration of any piece of information provided by the informant agents. Then, from all this knowledge, agents will be able to build different kinds of arguments, from which different kinds of conflicts may arise. Furthermore, agents will be equipped with a credibility order over informant agents, that may be used within an argument comparison criterion in order to decide between conflicting arguments. As a result, our approach extends Defeasible Logic Programming by enabling to argue for and against the consideration of informant agents, and resolving conflicts based on the credibility attributed to each information source, to finally determine the warranted beliefs the agents have.