Agents That Learn What Argument to Select In Argumentation-Based Negotiations

Abstract

Argument selection is considered the essence of the strategy in argumentation-based negotiation. An agent, which is arguing during a negotiation, has to decide what arguments are the best to persuade the opponent. In fact, in each negotiation step, the agent must select an argument from a set of candidate arguments by applying some selection criterion. For this task, the agent observes some factors of the negotiation context, for instance trust in the opponent, expected utility, among others. Usually, argument selection mechanisms are defined statically. However, as the negotiation context varies from a negotiation to another, defining a static selection mechanism it is not useful. For this reason, we present in this paper a novel approach to personalize argument selection mechanisms in the context of argumentation-based negotiation. The selection mechanism defines a set of preferences that determine how preferable it is to utter an argument in a given context. Our approach maintains a hierarchy of preferences in order to learn new preferences and update the existing ones as the agent experience increases. We tested this approach in a simulated multiagent system and obtained promising results.