Predictive Reasoning in Subjective Bayesian Networks

Abstract

Subjective Bayesian networks extend Bayesian networks by substituting the conditional probability distributions with subjective opinions. In that way they enable explicit representation of the uncertainty in the probabilistic information encoded in the network. In this paper we focus on predictive reasoning in subjective Bayesian networks and propose an inference method that is based on the operations of deduction and multiplication of subjective opinions. We demonstrate modelling and inference with subjective Bayesian networks through an example.