Resource Description Framework (RDF)

The Resource Description Framework (RDF) is a W3C standard data model for describing resources on the web. It was first issued as a recommendation in February 1999 (RDF Model and Syntax) and substantially revised as RDF 1.1 in 2014. Despite the name it is not a single file format; it is an abstract data model with multiple interchangeable serializations. Every RDF statement is a triple of subject, predicate, and object. The subject identifies a resource using a URI or a blank node. The predicate, also a URI, names a property or relationship. The object is either another resource (URI or blank node) or a literal value, optionally tagged with a datatype or language. A set of triples forms a directed labelled graph, and any two RDF graphs can be merged simply by taking the union of their triples, which is the property that makes RDF well suited to cross-dataset integration. RDF is the substrate for several layered standards. RDF Schema (RDFS) adds classes, subclass and subproperty hierarchies, and domain and range constraints. OWL, the Web Ontology Language, extends RDFS with richer logical constructs. SHACL provides validation rules. SKOS represents thesauri and taxonomies. Queries over RDF graphs are written in SPARQL, the W3C-standardized query language. Common serializations include the original XML-based RDF/XML; Turtle, a compact human-readable syntax; N-Triples, a minimal line-based format used for streaming and diffing; N-Quads and TriG, which extend the previous two to support named graphs; and JSON-LD, which embeds RDF in ordinary JSON and is widely deployed in HTML pages for search-engine structured data. RDF underpins much of the Linked Data and RDF ecosystem, including DBpedia, Wikidata, MusicBrainz, FOAF, schema.org-marked-up web pages, and many library-catalogue and cultural-heritage datasets. Its uptake outside those domains has been limited by the steeper learning curve compared with plain JSON APIs and property-graph databases.