In this paper, we give an overview of general classes of similarity measures which presents the advantage of being inserted in a cognitive framework and being available for objects described by means of non-classical kinds of values, such as linguistic values, as well as traditional numerical attributes. We also give means to prioritize one measure over the others.