The NBI software allows for processing of data available in practically any form, including the "yes - no" (1 or 0) parameters. For demonstration, we will refer to the example illustrated by Fig. 12 and 13. Fig. 13 represents a dissimilarity matrix of a graph of size 10. Fig. 8 (pink background section) shows the data analysis results produced by the NBI software in the form of informographic spectra based on the application of the information entropy function. On the plots, where information negentropy is shown as a function of the logarithm of the number of transformations of the target matrix, one can see a number of individual symmetrical peaks that are precisely described by the Gauss equation. In a spectrum for each object, the peaks with maximal retention numbers (RNm) are characteristic of primary clusters. In this case, one can see two primary clusters: A - ... - D, with RN = 65; and E - ... - J, with RN = 114. The obtained informospectra (informograms) suggest that the two primary clusters are separated at the very start of the transformation process, after which the hierarchical resolution proceeds independently within each of the primary clusters. For instance, in the second primary cluster, the pair of objects E + F (RN = 6) is the first one isolated, then follow G (RN = 15), H (RN = 22), I (RN = 41), and, finally, J (RNm = 114) - which lies in the base of the second cluster tree.

Notwithstanding the fact that transformation of the similarity matrix starts with formation of two primary clusters A - ... - D and E - ...- J, which is followed by subsequent resolution taking place within each cluster, the clustering process is performed based on a premise that all of the objects of the matrix as interrelated. A change of the value of even one parameter in the matrix - in this example, parameter F-I changes from 0 to 1 (see the cell with a double-line border) - results in drastic changes in informospectra of all the objects (Fig. 13, blue background section).

We define the described phenomenon as resolution of bulk information, occurring in a way similar to chromatographic resolution. There is practically a full analogy, including that performance peaks can be assessed based on a number of theoretical plates. Having said that, we suggest the definition of information as that which has no mass but is sorbable. In the reasoning model, based on which the NBI-algorithm has been designed, a "mobile phase" is the quasi-continuous process of matrix transformation, and a "stationary phase" is the matrix in which similarity values equal zero.