Rival similarity as an universal basic tool of cognitive data mining

During Data Mining tasks solving a person use a specific universal psycho and physiological cognitive technique. Key points of the technique are in a way of estimating a measure of similarity between objects and in necessity to maximize compactness and simplicity of a world description according to the measure. Nikolay Zagoruiko offered a measure of similarity, which takes into account a rival environment. In the paper the Function of Rival Similarity (FRiS-function) and some possibilities of its usage for patterns compactness and separability estimating are presented. These estimations are used in algorithms for solving classification (FRiS-Stolp algorithm), feature selection (FRiS-GRAD algorithm) and censoring (FRiS-Censor algorithm) tasks. Main ideas, some properties of the algorithms and their results on model and real tasks of data mining are described in the paper as well.