Pidnebesna H. Ontological models for intellectualization of the process of designing inductive modelling problem solution tools

The dissertation is devoted to the development of ontological models of the domain of inductive modeling based on GMDH for use in the designing of tools for modeling complex objects based on statistics. The ontology allows the automated design of rules for selecting effective methods for solving specific applications of modeling complex objects based on statistical data. This allows increasing the efficiency of developing the software for observation-based modeling. In the thesis, ontologies of inductive modeling domains are used as a formal structure of knowledge representation in tool development. This provides an increase in the level of intelligence of computer modeling systems. The methodology of the construction of ontological models of IM-area is presented. For the first time, ontological models of the main components of the GMDH-based modeling process were built. A new correlation-rating algorithm GMDH, which belongs to the class of directed search methods, is developed and investigated, and its efficiency is shown. The proposed algorithm is designed to solve the current problem of differentiating informative and non-informative factors. The developed methods are used to solve environmental and economic applications.