ECAI 2004 Conference Paper

[full paper]

Defining Equivalence Classes for the Elicitation of Probability Constraints for Bayesian Networks

Linda C. van der Gaag, Eveline M. Helsper

Among the tasks involved in building a Bayesian network for a real-life application, the task of eliciting all probabilities required is generally considered the most daunting. This task can be supported by the construction of a qualitative network as an intermediate model of the domain under study. Such a network specifies qualitative features of the probability distribution to be represented, which can be taken as constraints on the probabilities for the Bayesian network. In this paper, we analyse the possible combinations of features, resulting in a small number of equivalence classes. Based upon these classes, we present a method for eliciting the qualitative features of a domain's probability distribution. We report on an initial study of the use of our method in the domain of neonatology.

Keywords: Bayesian networks, Probabilistic relationships, Knowledge acquisition

Citation: Linda C. van der Gaag, Eveline M. Helsper: Defining Equivalence Classes for the Elicitation of Probability Constraints for Bayesian Networks. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.1103-1104.