Identification of Integrated Rating Mechanisms Based on Training Set

A problem of identification of the integrated rating mechanisms based on training set is one of the urgent mathematical problems in organizational behavior control science. The integrated rating mechanism consists of a binary tree, set of convolution matrices and scoring scales. Nowadays the one-hot approach applying for its solution. This approach based on the representation of the integrated rating mechanism in the form of a superposition of the products of one-hot encoded criteria and matrices. The identification of the criteria tree structure and elements of convolution matrices made by logical methods and polynomials like machine learning, and also algebraic methods - varieties. However, according to the known approach a training set elements present by discrete values. This paper presents an alternative approach to the identification of integrated rating mechanisms based on training set. It is shown that matrix rating mechanisms can be represented in the form of an artificial neural network. This circumstance opens up wide opportunities for the synthesis of identification methods for matrix rating mechanisms and neural network technologies for data processing.