Petri Net vs. Event Bush in Description of Environments of Alternative Directed Changes

Introduction

Knowledge base is an essential part of any expert system. When formulated as a set of rules, it gives ground for the inference, which draws conclusions from data entered by the user (Giarratano & Riley, 2005). This type of inference is specific for expert systems and provides advice and decision support (see, e.g., Arendt & van Uden, 2011) as well as systematic capture, transfer and reuse of knowledge (Quintana-Amate et al., 2015). Still, in contrast to databases that have various and well-developed standards of data structuring (e.g., those developed in relational or object-oriented framework), the knowledge base, in common practice, until now often represents a simple repository “containing articles that relate to fault fixing, how-to documents, user guides, white papers and so on” with an ad-hoc retrieval system (Serio Ltd., 2014). However, even when organized as a set of rules, or productions, the knowledge base often lacks proper tools of knowledge engineering to adequately incorporate the human expertise and account for the peculiarities of meanings. In fact, the only knowledge-engineering approach that is being used is ontology (taxonomy, hierarchy, classification)-related.

The examples of this approach in knowledge base design include, but are not exhausted by, ontologies and hierarchies per se (Kulon et al., 2006), structural models in the DANE method (Goel et al., 2012; DANE 2.0 Users Guide, 2011), UML class diagrams (La Rocca, 2012; da Silva et al., 2014), class and object frames (La Rocca, 2012), IDL code (La Rocca, 2012), “holistic models” (Syairudin et al., 2015), or “view models” of Akasaka et al. (2012) that allow to perform analogical reasoning and generate new functions of modeled systems. Product tree (La Rocca, 2012) is another ontology-related approach that resembles event tree but its “parent-child” relation maps not the “cause-effect” but the “whole-part” relation, thus performing an instantiation of a product model. “Customer satisfaction tree” (Akasaka et al., 2012) has inverse composition but maps the same relation, parent corresponding to part and child, to whole. Tree-shaped conceptual design graphs and their modifications presented by Calkins et al. (2000) are also based on part-whole relation and therefore represent another example of the same ontology-related approach.

Nevertheless, ontologies, conceptual graphs and other traditional tools of knowledge engineering/representation operate with objects considered either as classes (i.e., types: concept types or relation types) or as individuals (Martin, 2002). This has proven to be efficient for a wide range of tasks like building data models or making classifications and descriptions of those domains in which things, their properties and relations are considered to be fixed – e.g., production units of a factory or biological species. Nevertheless, there are some tasks in natural-scientific and technical domains that require consideration of changing properties and relations and claim for formal description of scenarios and patterns of evolution. The knowledge relating to such tasks has been termed dynamic, the methods of its processing, dynamic knowledge representation (engineering, management and so forth), and the contexts that include them, dynamic environments (Pshenichny & Kanzheleva, 2011).