Identification of Control Structures. Inductive Inference Methods for Controlled Rule Based Systems

摘要

The authors consider rule based systems with an explicit representation of control structures. Using such a controlled rule based system whose set of rules is assumed to be known, they address the problem of inferring a control structure from a set of traces produced by the system. Identification of such a control structure may be important in the building of expert systems or the simplification of such systems. Given a finite class of control structures and distinguishing three identification models (minimal identification, arbitrary identification behaviorally-correct identification), they show that the identification methods to be used in these models can be derived from a simple schematic algorithm. (Copyright (c) 1987 by Faculty of Technical Mathematics and Informatics, Delft, The Netherlands.)