The paper presents an approach for automatically extracting views from a relational database schema, based on the knowledge of the application domain. In order to achieve such result, the possible contexts emerging from the application domain are identified and, starting from the representation of the characterizing dimensions, a sequence of operations is applied to the schema to derive the portion of interest or, as we say, to tailor the schema with respect to each possible context. We introduce the tailoring operators as well as the method to compose partial views to obtain the final ones; a running example is used to lead the reader through the steps of the proposed methodology.