Abstract : RDF is a framework to publish statements on the web about anything. It allows anyone to describe resources, in particular Web resources, such as the author, creation date, subject, and copyright of an image. Any information portal or data-based web site can be interested in using the graph model of RDF to open its silos of data about persons, documents, events, products, services, places etc. RDF reuses the web approach to identify resources (URI) and to allow one to explicitly represent any relationship between two resources. Such statements can come from any source on the web and be merged with other statements supporting worldwide data integration. Using and reusing URIs, anyone can say anything about any topic, anyone can add to it, and so on. Additionally, using RDFS, one can define domain-specific classes and properties to describe these resources and organize them in hierarchies. These schemas are also published and exchanged in RDF. RDF not only provides a graph model to publish and link data on the web, it also provides the foundational shared data model on which other capabilities are built: querying (SPARQL is built on top of RDF), embedding (RDFa and GRDDL rely on the RDF model), and reasoning (RDFS and OWL are defined on top of RDF). Semantic web is a web to link data and share the semantics of their schemas. RDF provides a recommendation to publish and link data. RDFS provides a recommendation to share the semantics of their schemas. The couple RDF & RDFS is also reused in several other activities of the W3C.