Knowledge Verification with an Enhanced High-Level Petri-Net Model

PETRI NETS, WHICH CAN ANALYZE systems involving conditional relationships, provide straightforward models for rulebased systems.1 Researchers have used Petrinet models to study knowledge validation, verification, and testing.2–4 However, these models considered a rule-based system to consist of rules that are explicitly described; that is, no variables appear in the rule bodies. Moreover, these models fail to describe the relationships involving negative information in rules. Other researchers have investigated variables and negation.5–7 However, the closedworld assumption interprets the meaning of negation in rule-based systems, and these models do not describe such relationships exactly. Also, knowledge verification and validation in such models needs to be explored. In this article, we describe how to model rule-based systems using advanced Petri-net models in which variables and negation are exactly represented. We also explain how such models achieve knowledge verification. We explore the detection of improper knowledge, including redundancy, subsumption, conflicts, cycles, and unnecessary conditions, through reachability problems solved using an enhanced high-level Petri-net (Ehlpn) model. In this article, we assume that each rule describes an implication relation from a collection of conditions (left-hand side) to a collection of actions or conclusions (righthand side). Negative elements may appear in both the leftand right-hand sides. For convenience, we represent a variable in a rule by a string beginning with an uppercase letter, and a constant by a string beginning with a lowercase letter.