The problem of modeling uncertainty and inexact reasoning inrule-based expert systems is challenging on nonnative as well oncognitive grounds. First, the modular structure of the rule-basedarchitecture does not lend itself to standard Bayesianinference techniques. Second, there is no consensus...