The main goal of this paper is to evaluate knowledge base schemas, modeled as a set of entity types, each such type being associated with a set of properties, according to their focus. We intuitively model the notion of focus as ''the state or quality of being relevant in storing and retrieving information''. This definition of focus is adapted from the notion of ''categorization purpose'', as first defined in cognitive psychology, thus giving us a high level of understandability on the side of users. In turn, this notion is formalized based on a set of knowledge metrics that, for any given focus, rank knowledge base schemas according to their quality. We apply the proposed methodology to more than 200 state-of-the-art knowledge base schemas. The experimental results show the utility of our approach