Clusterization of objects with fuzzy parameter's values

We suggested a method of clustering, which allows to build a model of conceptual clustering for objects of fuzzy nature, and also to increase the accuracy of clustering for such objects. We used Cobweb clustering method as a base. We modified the formula of assessing the utility of conceptual clustering for objects with fuzzy parameter values. Then we suggested a modified Cobweb version for working with such objects. A numerical method for getting a piecewise linear and U-shaped membership functions for the parameters of clustered objects is developed. We implemented clustering framework for the objects with fuzzy parameter's values. Then we decided the task of user roles forming automation with using suggested method.