RAM: a Petri net based expert system with memory

A new method of expert systems (ES) based on Petri nets, called RAM, is presented. This is the first such system which can memorize the truth value of the proposition processed. Using RAM: 1) the representation of knowledge processing system using conventional Petri net can be developed; 2) both forward- and backward-chaining can be modeled; 3) the truth value of the same proposition is uniquely stored and hence only single copy of propositions is needed; 4) object oriented concept is used; 5) sufficient and correct representations of both Horn- and non-Horn-clauses with negations appearing in both the head and body of any production rule can be achieved; 6) theorem proving by refutation can be performed; 7) property verification for correctness, consistency, and completeness with polynomial complexity can be achieved; 8) activation token is not needed; 9) neither Petri-like net nor colored Petri net is necessary for the ES; 10) the condition under which the representation of ES can be simplified to a simpler ordinary Petri net is studied; and 11) dynamic simulation of the execution of ES can be conducted