Method of search and logical inference of expert information in a directed cyclic knowledge multigraph

Anatoly F. Zaytsev

East Siberian state university of technology and management

Currently, many modern problems of applied mathematics and computer science are solved using graph theory. Various complex systems, such as neural networks or knowledge bases, can be represented and described as graphs. The most common problems using graph theory are: finding the shortest path, determining the maximum flow in the network, finding the minimum spanning trees and others. At the same time, there are quite a lot of unresolved problems. The relevance of the work is due to increased interest in the fields of artificial intelligence and knowledge engineering, the methods of which are the possibility of transforming the obtained subject models into logical and mathematical, in the form of programs for computers that carry out computer or simulation modeling of the systems under study. In the presented work the process of development of the special mathematical and algorithmic software for system of the analysis and processing of the expert information for the purpose of the automated search and modeling of decisions of problems for identification of dynamic systems is described. The problem of search and logical inference of synthesized solutions of problems on knowledge graphs is formulated. The model of knowledge base of the selected subject area in the form of knowledge multigraph is presented, as well as the method of search and logical inference of solutions of problems, with their software implementation in the programming language Python. The presented method of search of weakly connected subgraphs and synthesis of problem solutions is implemented by using theoreticalmultiple analysis, as well as elements of graph theory. To demonstrate the performance of the method, its implementation in the form of an algorithm in the Python programming language and the results of computational experiments are given. The novelty and practical significance of the work lies in the fact that the proposed method and algorithms can be used in the practical implementation of knowledge bases, logical inference mechanisms and processing of expert information in a variety of expert, calculation-logical and hybrid intelligent systems, replacing their foreign analogues.

system analysis, logical inference, graph search, multigraph, expert system, knowledge base, knowledge graph, Python