Semantic Web

In 1989, Tim Berners-Lee, the future founder of W3C, devised an innovative organization of internal resources for CERN in the form of a distributed computer network. The proposed structure is that of a graph whose nodes represent resources (people, groups of people, projects, concepts, documents, hardware objects, entities of various kinds) and whose arcs represent labeled links connecting the nodes. Examples of labels proposed in 1989 are: depends on, is part of, realizes, refers to, uses, is an example of... This vision, based on the notion of hypertext, is at the origin of the Web. However, despite its rapid expansion, with an impressive number of sites and users, today's Web only imperfectly reflects the original vision. The Web only connects documents (untyped nodes), and the links themselves have no typed tags. Human users have learned to cope with these limitations by relying on their understanding of the content of connected documents and, where possible, assigning meaning to links using anchor text. However, these capabilities remain largely inaccessible to software in the absence of machine-interpretable semantics.

The Semantic Web (WS) initiative, spearheaded by the W3C, proposes a qualitative leap from the current Web that, in a way, echoes Berners-Lee's original vision. To quote Berners-Lee et al. in The Semantic Web, "the Semantic Web is an extension of the current Web in which information is provided with a well-defined meaning, enabling computers and people to work better together".

WS promoters propose the systematic construction of an additional level of explicit description of resources and links, for better use by machines. This specific level of representation is made up of what is often referred to as metadata or semantic annotations. It is based on a simple graph model, the RDF model (see § 2.1 ), which enables resources to be linked semantically. But it can also be used for complex representations. In the following, resource means anything that can be identified by a URI (Uniform Resource Identifier). These URIs have two roles: the first is identification, the second is addressing (linked to a protocol in the case of URLs), and so, in this case, location and access. So, resource, data in the broadest sense, means a book or a document, a geographical location, a concept in a conceptual schema, but...