Huginn: Normative reasoning based on mood

Abstract

A multi-agent system can be helped by a normative system that guides its (autonomous) agents towards an expected behavior. The agents on their side have to reason about the impact of those norms in its personal goals. Considering that agents have limited resources, it is necessary to reason also about available resources and whether they are enough to reach goals related to obligations. A proposal for this kind of reasoning is relevant in some applications, however current proposals for normative reasoning do not consider the limited resources. This paper proposes a deliberation process that uses the concept of mood to reason about norms, desires and resources. The proposed deliberation process translates the norms, desires and resources to an optimization problem known as multidimensional knapsack problem with multiple-choice.