Fast reasoning in a rule-based system with uncertainty

The knowledge processed in empirical domains is more or less uncertain. In order to support people who deal with them, expert systems with uncertainty are used. The expert systems that serve for planning or simulation purposes are often implemented as rule-based systems. To express the uncertainty of facts and rules, different mathematical methods are used: from probability factors and modal logics to the Zadeh's fuzzy logic. The last method is the most general, and it helps to conclude very reliable hypotheses. In the simulation systems both the conclusions' reliability and the time necessary for reasonings are of great importance. In this paper we point at the rule of convergence as a method of reasoning which allows to speed up reasonings performed in rule-based systems with uncertainty. We discuss its advantages, limitations and possible applications.