On the Detection of Con icts in Diagnostic Bayesian Networks Using Abstraction

An important issue in the use of expert sys tems is the so called brittleness problem Ex pert systems model only a limited part of the world While the explicit management of uncertainty in expert systems mitigates the brittleness problem it is still possible for a system to be used unwittingly in ways that the system is not prepared to address Such a situation may be detected by the method of straw models rst presented by Jensen et al and later generalized and justi ed by Laskey We describe an algorithm which we have implemented that takes as input an annotated diagnostic Bayesian net work the base model and constructs with out assistance a bipartite network to be used as a straw model We show that in some cases this straw model is better that the in dependent straw model of Jensen et al the only other straw model for which a construc tion algorithm has been designed and imple mented