In this article the problem of automatic generation of a knowledge base which consists of production rules for training dataset using fuzzy logic methods and a rule for comparing the values of an output variable is considered. An algorithm for the formation of fuzzy production rules is proposed. An actual problem of development and improvement of artificial intelligence algorithms and fuzzy logic application for solving a wider range of problems is considered. With the help of such systems are possible to eliminate the difficulties of formalizing knowledge about technological processes; also it is possible to organize recognition of nonstandard and emergency situations without using precise mathematical models and classical decision theory based on the tool of mathematical equations. The development of this area is relevant, as the number of tasks are constantly increasing, and the amount of knowledge becomes too large to handle them manually. The construction of an exact mathematical model for poorly formalized objects and processes are very difficult task, due to the lack of complete information. The situation becomes even more complicated if the properties of the object or process change dynamically. Therefore, the development of mathematical methods and algorithms that allow structuring the system of rules and determining the order of their calls to control consistency and completeness to optimize the number of rules, are an actual task. Modern approaches to the automation of these processes are considered. These approaches significantly improve the work of expert systems, but they allow to work only with static knowledge bases, limit the number of logical inferences and are not applicable for cases when it is necessary to add new logical rules to the existing system. In this article, an approach is developed that makes it possible to expand the knowledge base of the expert system with new rules in the process of exploitation. The developed algorithm has following advantages: high speed of problem solving; the ability that allows expanding the number of system responses without changing the scope of the rules and the program itself; expanding the range of application of fuzzy logic algorithms. The developed algorithm has following disadvantages: if the system's response database has objects that are similar to each other, they can have the same center of gravity, which in turn leads to additional checks; the minimum distance for mapping the object should be selected experimentally. The application of this algorithm can be seen on the website of the program, which classifies, maps an arbitrary user in a set of comic book characters database "CMD - Combat Marvel DC" [8]. The approach that was proposed has been successfully implemented using the C/C ++ and JavaScript languages, and JSON open-standard file format that uses human-readable text to transmit data objects consisting of attribute–value pairs and array data types. Software that was used for development: NetBeans IDE, MinGW, GNU Compiler Collection, WhiteStarUML, GitHub, WebGL, Chrome, Mozilla Firefox, Opera
Ontological approach to automatic generation of questions in intelligent learning systems
