The Semantic Web aims at easy integration and usage of content by building on a semi-structured data model where data semantics are explicitly specified through ontologies. The use of ontologies in real-world applications such as community portals has shown that a new level of data independence is required for ontology-based applications. For example, the customization of information towards the needs of specific user communities is often need. This paper extends previous work [22, 21] on this issue and presents a view language for the fundamental data models of the Semantic Web, viz. RDF and RDFS, and how it can be implemented. The basic novelty of the view language is the semantically appropriate classification of views into inheritance taxonomies based on query semantics. Additionally, the underlying distinction between unary predicates (classes) and binary predicates (properties) taken in RDF/S is maintained in the view language. So-called external ontologies allow the integration of multiple source databases, offer control over the publishing of data and enable the generation of views spanning across databases.