Methods of rules selection with backward chaining in static expert systems

The article discusses the problem of optimizing the solution search process in a static expert system. The research on the stage of backward chaining for rule selection and execution is conducted. A method of statistics collection for analysis of this stage is given. The results are presented in the form of a comparative analysis of the four methods of selecting rules for knowledge bases in three categories of tasks. The most efficient methods are identified in each category. Based on these methods, a version of the mixed method and an algorithm of backward chaining minimizing the number of rules used in solving the tasks under study are proposed.