A construction of a compressed description of data using a function of rival similarity
Н. Г. Загоруйко, Irina A. Borisova, Olga A. Kutnenko, Vladimir V. Dyubanov · Journal of Applied and Industrial Mathematics · 2013
We justify the claim that all kinds of data mining aim at constructing compressed and simplified descriptions of information. We propose passing from the binary absolute measure of similarity between objects to a ternary relative measure: a function of rival similarity (FRiS-function). Its use enables us to obtain a quantitative estimate for the compactness of data and construct new, more effective cognitive analysis methods. We present examples of solutions to various model and real problems by the new methods.