Abstract
This paper introduces a notion of relevance (conceptual stability) of information granules. Granulation of data results in a series of chunks of information usually referred to as information granules. These information granules are basic building entities involved in the design of a broad class of systems. Information granules are also percepts - entities being perceived by humans as being essential while working with some real-world phenomena, especially describing and interacting with them. The percepts need to be comprehensible. They should also reflect the experimental evidence. Furthermore, information granules should be stable, meaning that they reconcile experimental reality with the subjective and ultimately observer-based judgement about the environment. Once being stable, information granules could be viewed as architecture-independent. The proposed algorithmic environment supporting this concept dwells on the ideas of statistical inference that helps quantify stability thorough nonparametric testing.