Abstract
Local sources of knowledge structured in the form of logic descriptors-constructs of fuzzy logic, are arranged together (structured) in the form of a global model coming as a high-level granular logic descriptor. The inherent granularity of the global descriptor of this nature arises as a manifestation of the diversity of the locally available descriptors. The granular descriptor can be expressed with the aid of any of the formal models of information granules including sets, fuzzy sets, rough sets, probabilistic granules and others. The architectural essence of the granular descriptor, which supports a quantification of the variability among the sources of knowledge, is realized through an optimal allocation of information granularity. Information granularity is treated as an important design asset and its allocation throughout the parameters of the logic descriptors helps quantify the diversity of individual sources of knowledge. Various protocols of allocation of information granularity along with an overall quantification of their effectiveness are discussed along with their numeric characterization.