Standards · Serialisation

JSON-LD: linked data in the format developers already use

JSON-LD adds a context to ordinary JSON, so documents an application already exchanges can carry linked-data meaning without a new parser or pipeline.

Standards · Turtle · GraphQL-LD · Knowledge formats

What it is

JSON-LD is a JSON-based format for serialising linked data: it keeps JSON's syntax and adds a few keywords that say what the keys mean. The context maps each term to an IRI, @id identifies a node, and @type states what it is.

What it gives the knowledge layer

It gives the knowledge layer a way to reach systems whose owners will not adopt a triple syntax. The meaning travels inside the payload, so an API already returning JSON can return linked data, and one stored graph can be served in the shape each client expects.

Where the spec lives

JSON-LD 1.1 is a W3C Recommendation published in July 2020, one of three documents: the syntax, the Processing Algorithms and API, and Framing. The graph it serialises is defined by RDF 1.1 Concepts, the model behind Turtle and the target of SPARQL.

What it does not solve

The context is the whole game. A term means whatever the context binds it to, so a wrong or drifting context silently changes the meaning of every document using it, and a remote context can change without those documents changing. Protected term definitions reduce that risk; they do not remove it.

FAQ

JSON-LD is JSON plus a context. The context connects a term such as name to the IRI it stands for; without it the keys are just strings two systems may read differently.

They transform between document forms: compaction shortens IRIs against a context, flattening lifts embedded nodes out and replaces them with references, and framing reshapes the result against an example frame.

Because it is the only thing tying a term to an IRI. Change it and the meaning of every document using it changes too; a remote context can change without the documents changing.

They serialise the same RDF graph and suit different hands: JSON-LD where the pipeline already speaks JSON, Turtle where a person authors and reviews the model. Neither changes what the triples mean.

Sources

  1. JSON-LD 1.1 — A JSON-based Serialization for Linked Data (W3C Recommendation, 2020)
  2. RDF 1.1 Concepts and Abstract Syntax (W3C Recommendation)