On Ranking Production Rules for Rule-Based Systems with Uncertainty

There are many places (e.g. hospital emergency rooms) where reliable diagnostic systems might support people in their work. The paper discusses the problem of designing diagnostic rule-based systems with uncertainty. Most such systems use the technique of forward chaining in their reasonings. The number and the contents of the hypotheses depend then on both the form of system's knowledge base and the details of the inference engine performance. In particular, the hypotheses can be influenced by the rules' priorities. In the paper we propose a method for determining priorities for the rules designed from true evidence base which contains aggregate data of an attributive representation.