RDF Schema (RDFS) and RDF are evolving as the de facto languages for the Semantic Web. Whereas RDFS can be used to define concepts in a domain ontology, RDF can be used to define instances of that ontology. In an attempt to make these modeling languages evolvable and extensible, the current languages lack a clear semantic support. While RDF and RDFS are easy to use, they are hard to understand and difficult to validate and extend as a result of their ambiguous semantics. In this paper, we use UML modeling and meta modeling concepts to resolve some of semantic ambiguities and provide support for validation and extension. We describe a UML meta modeling approach that defines five modeling layers and identify how RDFS and RDF as Semantic Web languages fit into the architecture. The approach also defines the meta languages required to create RDFS and RDF and to support extensions to both languages and their constructs. The meta modeling approach is based on concepts inspired by the UML meta modeling architecture. Moreover, we use UML to bootstrap the modeling hierarchy and we use the UML extension mechanisms to describe the models developed at each layer including the RDFS and RDF language constructs. 19 Pages