Knowledge represented in the semantic network, especially in the Semantic Web, can be expressed in attributive language AL. Expressions of this language are interpreted in different theories of information granules: set theory, probability theory, possible data sets in the evidence systems, shadowed sets, fuzzy sets or rough sets. In order to unify the interpretations of expressions for different theories, it is assumed that expressions of the AL language can be interpreted in a chosen relational system called a granule system. In this paper, it is proposed to use information granule database and it is also demonstrated that this database can be induced by the measurement system of the adequacy of information retrieval, called a perceptual system. It can simplify previous formal description of the information granule system significantly. This paper also shows some examples of inducing rough and fuzzy granule databases by some perceptual systems.