TriX

TriX (Triples in XML) is an XML serialization for RDF that was designed to represent named graphs — that is, sets of RDF triples each associated with a graph name, giving an effective quad model of subject, predicate, object, and graph. It was introduced in 2004 by Jeremy Carroll (HP Labs) and Patrick Stickler (Nokia) in the paper "RDF Triples in XML," as a companion to the theoretical Named Graphs work that also motivated the more Turtle-like TriG. Unlike RDF/XML, which optimizes for terse, human-authored prose and offers many syntactic shortcuts, TriX deliberately favors a highly regular, minimal, and predictable XML structure.

Structurally a TriX document is a root TriX element in the namespace http://www.w3.org/2004/03/trix/trix-1/, containing one or more graph elements. Each graph may carry a name (a uri or blank-node id) and holds a flat sequence of triple elements, each with exactly three child terms in subject-predicate-object order. Terms are expressed with a small closed vocabulary: uri for IRIs, id for blank nodes, plainLiteral for string literals (with an optional xml:lang attribute), and typedLiteral for datatyped values (with a datatype attribute pointing at an XSD or other type IRI). Because every RDF term maps to one explicit element, the structure is fully self-describing with no prefix expansion, list sugar, or nesting to interpret.

The design rationale is toolability: because TriX is plain, normalized XML, it can be produced, queried, transformed, and validated with off-the-shelf XML machinery such as XSLT, XQuery, and XML Schema, and the specification even defines an XSLT-based extensibility mechanism for shorthand syntaxes. This makes it attractive for pipelines that already live in the XML ecosystem and need to move RDF datasets through them.

In practice TriX occupies a niche. It never became a W3C Recommendation, and mainstream RDF interchange today gravitates toward Turtle, TriG, N-Triples, N-Quads, and JSON-LD. It remains supported as a read/write format in Eclipse RDF4J (formerly Sesame) via its Rio parsers, in Python's RDFLib (Python), and by RDF toolkits such as Jena/RIOT, so it endures mainly as a bridge for XML-centric processing and archival of quad-level data rather than as a primary serialization.