What Are Linked Data?

Linked data are a set of design principles and technologies for publishing and connecting structured information on the web so that it can be discovered and consumed by machines as well as people. Rather than storing data in isolated silos, linked data uses standard web identifiers and protocols to make individual data items addressable and to express explicit relationships between them across different datasets and organizational boundaries. The approach builds on the architecture of the World Wide Web and extends it from a web of documents to a web of data.

Tim Berners-Lee introduced the four principles of linked data in a 2006 W3C design note: use URIs to name things, use HTTP URIs so those names can be looked up, respond with useful information using standards such as RDF when a URI is accessed, and include links to other URIs so that more related data can be discovered. These principles, often called the "linked data star rating" in discussions of openness, provide a practical baseline for interoperability among independently maintained datasets.

Knowledge Representation and Ontologies

Linked data relies on the Resource Description Framework (RDF) as its core data model. RDF represents information as triples: a subject, a predicate, and an object. Each element is identified by a URI, which means any RDF triple can reference entities or properties defined in an external dataset. Ontologies, expressed in languages such as OWL (Web Ontology Language) or SKOS (Simple Knowledge Organization System), define the vocabulary of a domain and the logical relationships between concepts. When two datasets use the same ontology terms, or when owl:sameAs links explicitly equate their entities, a reasoner can infer facts that neither dataset contains on its own. Well-known ontologies such as Dublin Core for metadata, Schema.org for web content, and FOAF (Friend of a Friend) for social networks provide reusable building blocks that reduce the effort of cross-dataset integration.

SPARQL and Query Processing

SPARQL is the W3C-standardized query language for RDF data, defined in the SPARQL 1.1 specification maintained by the W3C. A SPARQL query expresses graph patterns against one or more RDF datasets, retrieving triples that match the specified subject-predicate-object combinations. Federated SPARQL queries extend this capability across multiple SPARQL endpoints, allowing a single query to retrieve and join data from geographically and organizationally distributed sources without replicating datasets locally. Triple stores such as Apache Jena, Virtuoso, and GraphDB are purpose-built storage and query engines for RDF data, optimized for the characteristic access patterns of linked data applications. As dataset sizes grow into the billions of triples, query optimization and graph indexing become critical concerns, connecting linked data to the broader field of large-scale knowledge graph processing.

Linked Open Data

The Linked Open Data (LOD) movement applies linked data principles to publicly accessible datasets, building a global cloud of interlinked information spanning government statistics, scientific data, cultural heritage collections, and geographic information. DBpedia, which extracts structured data from Wikipedia's infoboxes and publishes it as RDF, and Wikidata, a collaboratively maintained knowledge graph covering approximately 100 million items, are two of the largest and most heavily referenced nodes in the Linked Open Data cloud. Government portals such as data.gov and the European Data Portal have adopted linked data formats to expose public datasets in machine-readable form. The practice of attaching rich metadata to datasets, including provenance, licensing, and update frequency, enables automated discovery and reuse across organizational boundaries.

Applications

Linked data has applications across a wide range of data-intensive domains, including:

• Enterprise knowledge graphs for product catalogs, customer data, and supply chain information
• Biomedical research databases linking genes, proteins, diseases, and clinical trials
• Library and cultural heritage systems for cross-collection discovery
• E-commerce and search engine optimization through Schema.org structured markup
• Open government data portals enabling cross-agency analytics