Abstract
In light of the remarkable diversity of data, arises an interesting and challenging problem of their description and concise interpretation. In a nutshell, in the proposed description pursued in this study, we consider a framework of information granules. The study develops a general scheme composed of two functional phases: (i) clustering data and features forming segments of original data and delivering a meaningful partition of data, and (ii) development of information granules. In both phases, we discuss a suite of performance indexes quantifying the quality of segments of data and the resulting information granules. Along this line, discussed are collections of information granules and their mutual relationships. A series of publicly available data sets is used in the experiments-their granular signature is quantified, and the quality of these findings is analyzed.