My PhD focusses on argumentation-based communication between agents. I take, as a starting point, the argumentation system proposed by Garcia and Simari [Alejandro J. Garcia and Guillermo R. Simari. Defeasible logic programming an argumentative approach. Theory and Practice of Logic Programming, 4(1-2):95-138, 2004], which allows a single agent to reason about its beliefs. I define a novel dialogue system that allows two agents to use Garcia and Simari's system to carry out inter-agent argumentation. I define two specific protocols for two different types of inquiry dialogue that I define: argument inquiry and warrant inquiry. Argument inquiry dialogues are often embedded within warrant inquiry dialogues.
Other existing inquiry dialogue systems only model dialogues, meaning that they describe what a legal inquiry dialogue is, but they do not provide the means to actually generate such a dialogue. Such systems provide a protocol, which dictates what the possible legal next moves are at each point in a dialogue but not which of these moves to make. I present a system that not only includes two dialogue game style protocols, one for the argument inquiry dialogue and one for the warrant inquiry dialogue, but also includes an intelligent strategy, for an agent to use with these protocols, that selects exactly one of the legal moves to make.
As my system is generative, it allows me to investigate the precise behaviour of the dialogues it produces. I propose a benchmark against which I compare my dialogues, and use this to define soundness and completeness properties for argument inquiry and warrant inquiry dialogues. I show that these properties hold for all dialogues produced by my system. Finally, I go on to define another intelligent strategy for use with warrant inquiry dialogues. I show that this also leads to sound and complete dialogues but, in many situations, reduces the redundancy seen in the dialectical tree produced during the dialogue.
| 