Colored Knowledge Petri Nets and Logical Inference

Abstract

It is difficult to visualize and comprehend a knowledge Petri net since its size polynomially grows with the number of logical symbols. In order to tackle the issue, a colored-Petrinet based approach is proposed to perform the logical inference. First, a correlative symbol set is defined to represent the set of symbols that have relations via sentences in a knowledge base, and the semantic constraints are similarly categorized into several correlative constraint sets. Second, the generalized symbols and generalized semantic constraints are defined according to the coefficient matrix of a semantic constraint set, which have the same constraint on multiple sets of correlative symbols. Third, an algorithm is constructed to design a colored knowledge Petri net based on generalized symbols and generalized semantic constraints. Fourth, an inference engine is proposed based on the newly defined transition-firing rules, and can be used to infer or reveal hidden facts. The wumpus world problem is taken as an example to illustrate and verify the proposed method.