Resource Description Framework (RDF)

RDF models statements as triples: a subject (an IRI or blank node), a predicate (an IRI denoting a property), and an object (an IRI, literal, or blank node). This graph-based structure allows merging data from multiple sources without a fixed schema, a key advantage over relational tables.¹

RDF can be serialized in several syntaxes. RDF/XML was the original W3C recommendation, but Turtle (Terse RDF Triple Language) is now widely preferred for readability. Other formats include N-Triples, JSON-LD, and RDFa for embedding in HTML. Each syntax expresses the same abstract graph, enabling interoperability across tools.²

RDF provides a formal semantics that allows machines to derive new triples from existing ones. The RDF Schema (RDFS) adds vocabulary for classes, properties, and hierarchies, enabling basic inference such as subclass transitivity. OWL (Web Ontology Language) extends this with richer logical constructs like cardinality and equivalence.³

Inference engines, or reasoners, apply these rules to expand the graph. For example, if cat is a subclass of animal and Felix is a cat, a reasoner infers that Felix is an animal. This capability underpins data integration and knowledge graph construction, though it raises computational complexity trade-offs.⁴

RDF is used in diverse domains. Linked Data principles, proposed by Tim Berners-Lee, rely on RDF to publish and connect structured data on the Web. Major datasets like DBpedia and Wikidata expose their content as RDF, enabling cross-dataset queries via SPARQL endpoints.⁵

In the life sciences, RDF powers initiatives like Bio2RDF and the Open PHACTS platform, integrating genomic, chemical, and pharmacological data. Cultural heritage institutions use RDF for metadata standards like CIDOC-CRM. Even social media platforms, such as Facebook's Open Graph protocol, are built on RDF concepts.⁶

RDF has several niche features. Reification allows making statements about statements, but it is often criticized for complexity and is rarely used in practice. Named graphs extend RDF to manage provenance by grouping triples into distinct graphs, a feature formalized in RDF 1.1.¹

Blank nodes, which have no global identifier, are a source of subtle issues in graph merging and querying. RDF also supports datatype literals (e.g., integers, dates) and language-tagged strings, enabling multilingual data. The RDF-star extension, still under development, aims to simplify reification by allowing triples as subjects or objects.⁷

Triple

A statement in RDF consisting of subject, predicate, and object.

IRI

Internationalized Resource Identifier, a generalization of URI used to name resources.

Blank node

A node without a global identifier, used for anonymous resources.

SPARQL

A query language and protocol for RDF data.

OWL

Web Ontology Language, a knowledge representation language for defining ontologies.