Heuristic method of generalization of fuzzy productions

DOI: 10.31673/2412-4338.2018.044046

Abstract

The generalization process is based on the comparison of descriptions of the original objects, given by a set of characteristic values, and the selection of the most characteristic fragments of these descriptions. This process is also called inductive formation of concept. The task of the inductive formation of concepts, also called the task of generalizing concepts by means of attributes, is to build a concept based on the analysis of a sample, which allows using a certain recognition rule to correctly separate all positive and negative objects of the set of the sample under consideration. This paper addresses the problem of inductive generalization of a knowledge base of a production type. An important requirement should be, on the one hand, preserving the knowledge offered by experts, and, on the other hand, reducing the total volume of the knowledge base. In addition, the possibility of choosing the most likely allegations from the base must also be ensured. The paper presents a heuristic method for solving the problem of generalization of fuzzy productions. The method allows replacing a complete search of knowledge base productions with the faster procedure that takes into account the frequency of occurrence of facts in the productions of expert opinions. A method for presenting products in the form of Boolean functions has been proposed, with which the numerical weights have been assigned to control their processing. The stages of the generalization process have been developed. The results of the work can be used to build knowledge bases of expert systems in order to reduce the size of the knowledge base by inductive summarizing statements presented in the form of fuzzy productions, with obtaining an almost minimal set of productions excluding reciprocal fragments.

Keywords: expert system, fuzzy products, knowledge base, confidence coefficient, assurance rules.

References
1. Shatokhina N. K., Osmanow V. S. (2016). The Use of Probabilistic Genetic Algorithms with an Adaptive Mutation for the Problem Of Inductive Generalization of a Fuzzy Knowledge Base. Scientific works of the Donetsk State Technical University. Series: Informatics, Cubernetics and Computing. 2(23). 81-86.
2. Osmanow V. S. (2016). The Use of Genetic Algorithms for Constructing Inductive Generalization of a Fuzzy Knowledge Base. II International Scientific and Practical Conference "Modern Science: Problems and Prospects" (P. 2) October 15-16, 2016 - Kyiv ICSTD. 27-29.
3. Shatokhina N. K., Shatokhin P. A. (1999). On the inductive construction of the knowledge base of expert systems. Scientific works of the Donetsk State Technical University. Series: Computing and Automatics. 12. 158-164.
4. Grunsky I. С., Shatokhina N. K. (2001). On inductive generalization of fuzzy conclusions. Scientific works of the Donetsk State Technical University. Series: Computing and Automatics. 25: 154-160.
5. Fomenko T. N. (2018). Higher Mathematics. General algebra. Elements of a tensor algebra. Moscow: Publishing House Yurayt. 121.