Shabadash D. Methods of account of data's information granulation uncertainty at subject domain modelling in artificial intelligence systems.

The thesis is devoted to an increasing of informativity, reduction of description's dimension and size of subject domain in AI systems by the formalized formation of its model. In work concepts of hidden and shown uncertainty of data granulation were entered and quantity indicators for their esti-mation were developed. On the basis of these parameters the method of linguistic variable formation received further devel-opment. It is offered to formalize this process by estimation of variable's uncertainty level. Besides on the basis of the offered parameters was developed method of formalized formation of dimension. Also in the work the method of aggregates formation in the OLAP-systems, based on estimation of hidden uncertainty quantity of cube's granules was developed. The practical realization of the developed methods was executed in the form of program system Dimensions.