An approach for integrating statistical decision-support data with fuzzy action rules

This article describes an approach of integrating statistical decision-support information with fuzzy action rule systems, extending the concept of action rules towards the domain of probabilistic or statistical reasoning. Fuzzy action rules, inspired by fuzzy logic, allow describing the behavior of software agents in terms of linguistic variables. In principle, mapping statistical information into action rule systems allows implementing, e.g., advanced tutoring systems, whose behavior can be programmed by examples. As a consequence, these agents are also capable of learning upon experience, by allowing the action rules (the linguistic variables) to adapt, according to the perceived, actual applications.