A Multi-Agent Architecture for Decision Making under Uncertainty

An influence diagram represents a decision problem formalisation where the decision-maker is a rational one; i.e. he prefers the decision that maximises the utility function expected value. This formalism models the domain through a single homogeneous network and without considering the existence of specific sub-domains into the domain at hand. The main goal of this research project is to study the application of the multi-agent approach to the evaluation of influence diagrams, motivated by the intention of exploring the locality propriety in large domains. A domain that has locality can be looked by parts because it has specific and natural sub-domains. The decision-maker that reasons and decides based on the domain information can focus on a single sub-domain each time, and occasionally shift his attention to another sub-domain. This technical report was written when the first author was visiting the Computer Science Department of the University of Lisbon. It is the thesis proposal for the Computer Science doctor degree of the first author at the Brazilian Federal University of Rio Grande do Sul, and it was submitted with success last Apri