Multi-Player Belief Calculi: Models and Applications

Abstract

In developing methods for dealing with uncertainty in reasoning systems, itis important to consider the needs of the target applications. In particular,when the source of inferential uncertainty can be tracked to distributions ofexpert opinions, there might be different ways to model the representation andcombination of these opinions. In this paper we present the notion of multiplayerbelief calculi - a framework that takes into consideration not only the'regular' type of evidential uncertainty, but also the diversity of expert opinionswhen the evidence is held fixed. Using several applied examples, we show howthe basic framework can be naturally extended to support different applicationneeds and different sets of assumptions about the nature of the inference process.