Abstract
Information granules are conceptual entities using which experimental data are conveniently described and in the sequel their processing is realized at the higher level of abstraction. The central problem is concerned with the design of information granules. We advocate that a principle of justifiable granularity can be used as a sound vehicle to construct information granules so that they are (i) experimentally justifiable and (ii) semantically sound. We elaborate on the algorithmic details when forming information granules of type-1 and type-2. It is also stressed that the construction of information granule realized in this way follows a general paradigm of elevation of type of information granule, say numeric data (regarded as information granules of type-0) give rise to information granule of type-1 while experimental evidence coming as information granules of type-1 leads to the emergence of a single information granule of type-2. We discuss their direct applications to the area of system modeling, in particular showing how type-n information granules are used in the augmentation of numeric models.