Modeling production rules by means of predicate transition networks (Q1058869)

The purpose of the paper is that of trying to provide knowledge-based systems, using production rules as knowledge representation mechanism, with a formal tool suitable for proving properties, such as consistency or contradiction of the knowledge base. To this aim, the paper analyzes the possibility of encoding a knowledge base of production rules into a Predicate-Transition Network (PrTN). The model of PrTn was first introduced by \textit{H. J. Genrich} and \textit{K. Lautenbach} as a kind of high level Petri nets [Theor. Comput. Sci. 13, 109-136 (1981; Zbl 0454.68052)]. The peculiarity of this model lies in its well formalized semantics, which can be described by means of an algebraic formalism and supplies methods, such as that of S-invariants, for its analysis. The paper gives examples of typical constructs, occurring in production rules, represented with PrTNs and investigates the problem of the control (inference engine) of the network execution in relation to that of state reachability. An interesting research area to develop would be that of individuating subclasses of PrTns (meaningful for AI applications) for which the S-invariant method is both necessary and sufficient with respect to the reachability problem; in this case the method could be automatized in view of proving consistency and completeness of the net.