Development of Unified Approaches to Building Neural Network and Mathematical Models Based on Digital Data

Abstract

The paper considers the problem of developing approaches to building mathematical models based on digital data of real objects. The data are in text format and contains information about the behavior of the dynamic system. The information selected from the text data enables building of neural network and mathematical models of the dynamic system. The adequacy of the models is evaluated by analytical and numerical methods. The results are meaningfully interpreted. As a result of the study, it was confirmed that the algorithms and approaches for building mathematical models to solve the considered range of problems using digital data can be unified. The analysis of the obtained solutions showed that the con-clusions drawn on the basis of the built mathematical models and the conclusions drawn with the se-mantic neural network analysis of texts are consistent with each other. Therefore, one can talk about the positive results of the models developed. The models developed can be used in solving managerial tasks, planning and situation prediction.