Multilevel algorithms for precedent-type decision-making problems

In this paper, we considers a special class of precedent-type decision-making problems, which often arise in weakly formalised subject areas. To solve such problems, as a rule, heuristic algorithms are used, which cannot be strictly justified. It is shown that this class of problems can be reduced to a standard problem of pattern recognition with learning. Instead of heuristic algorithms, this allows to use multilevel models that make it possible to improve the accuracy of the solution, and in some cases to justify its correctness. An analysis of different variants for constructing multilevel models is given. A multilevel algorithm for the decision-making problem based on the structuring of information is proposed.