Abstract
Our concern here is in finding the objects in a database that have a desired value for a given attribute where our knowledge of the attribute value for the database objects is unclear. Here, the unclear attribute values are expressed using a generalized belief structure that contains both granular aspects and random aspects. Further, our target attribute value is also not clearly specified and is also expressed using a generalized belief structure but one with different granular components. We first discuss the properties of a monotonic measure and use this to define generalized belief structures, which we use to model our unclear information. We show that the degree of matching in this case of unclear database object values and unclear target values is interval valued which in makes the problem of finding the optimal database object difficult since we must compare interval values. Finally, we introduce the golden rule method for enabling a comparison between interval-valued degrees of matching.