Compendium

Semantic Web / RDF / Linked-Data / Ontology DSLs Family Index

15 min read

Semantic Web / RDF / Linked-Data DSLs Family Index

type: language-family-index family: semantic-web languages_catalogued: 30 tags: [language-reference, family-index, semantic-web, rdf, owl, shacl, shex, sparql, json-ld, turtle, rml, schema-org, linked-data]

Semantic Web / RDF / Linked-Data — Family Index

Family overview

The Semantic Web family is the cluster of W3C-track text languages that together implement the original Berners-Lee vision of a machine-readable web of data. The stack splits cleanly along four axes: a single abstract data model (RDF — subject/predicate/object triples), multiple concrete serialisations of that model (Turtle, TriG, N-Triples, N-Quads, RDF/XML, JSON-LD, RDFa), an ontology layer for describing classes and properties with formal semantics (RDFS, OWL 2 and its DL/EL/RL/QL profiles, SKOS), a shape / constraint layer for validating instance data against expected structure (SHACL, ShEx), a mapping layer that lifts non-RDF data into RDF (R2RML, RML, YARRRML), and a query layer (SPARQL 1.1, soon 1.2, with extensions like GeoSPARQL). They are not one language — they are a deliberately layered family with a shared abstract model, and that is why this index needs six tables rather than one.

The 1999–2010 era was a long stretch of vision exceeding delivery: RDF 1.0 (1999), OWL 1 (2004), SPARQL 1.0 (2008), and RDFa (2008) all shipped but adoption stayed niche, partly because RDF/XML was nearly unreadable and partly because the inference promises of OWL DL collided with web-scale practicality. The 2011–2014 reset changed the trajectory: RDF 1.1 (2014) made Turtle a first-class serialisation, JSON-LD 1.0 (2014) gave the web a JSON-friendly RDF that browsers and developers would actually adopt, and SPARQL 1.1 (2013) added update, federated query, and aggregation. Schema.org (launched 2011 by Google, Bing, Yahoo, later Yandex) quietly became the most-deployed RDF vocabulary on the planet by piggybacking on JSON-LD-in-<script> for SEO. SHACL 1.0 (2017) finally gave RDF a constraint language with industry consensus, ending years of ShEx-vs-SHACL competition.

The 2020–2026 period is the modern semantic-web 2.0 era, driven by three forces. First, RDF-star / SPARQL-star (originally Olaf Hartig, 2014, then a 2021 W3C Community Group Report) solved the long-standing pain of statements-about-statements (reification) by allowing a triple to appear as a term — and that work has now been merged into the core as RDF 1.2 and SPARQL 1.2, both currently at W3C Working Draft / Candidate Recommendation stage with target Recommendation no earlier than May 2026. Second, the knowledge-graph wave (Google’s Knowledge Graph, Wikidata, enterprise KGs from Bloomberg, eBay, Springer Nature, every major pharma) finally gave RDF its commercial vindication: it turns out the world wants persistent identifiers, schema-flexible joins, and SPARQL federation when it has a thousand heterogeneous sources to integrate. Third, the RML 1.0 (2024) and YARRRML toolchain matured into a credible “lift any CSV/JSON/XML/database into RDF” pipeline, and the LLM era has produced renewed appetite for KGs as grounding substrates for retrieval-augmented generation.

GeoSPARQL 1.1 (OGC 22-047, 2024) brought first-class support for spatial measurement, FeatureCollections, and SHACL validation profiles. SHACL 1.2 First Public Working Drafts (Core, SPARQL Extensions, Rules, Node Expressions) appeared in 2025–2026. Solid Protocol 0.12 (2025) is Tim Berners-Lee’s continuing attempt to use the RDF stack as the substrate for personal-data pods. The story is no longer “the semantic web that never was” — it is a quietly load-bearing layer of every modern data-integration platform, even when nobody markets it that way.

In our deep library

None of this family has standalone deep-library notes — they are all part of the W3C/OGC standards layer rather than first-class programming languages.

Cross-reference:

  • query — sibling family covering SPARQL alongside Cypher/Gremlin/GraphQL/etc. as general query languages. This family covers SPARQL specifically in the context of the RDF stack.
  • notation-spec — sibling family for formal-grammar/spec languages (EBNF, ANTLR grammars, ASN.1). Conceptually adjacent — OWL is a formal-spec language too, in a sense, but for ontologies rather than syntax.
  • citation-formats — Schema.org’s ScholarlyArticle / Dataset / CreativeWork profile is the dominant web-discoverability vocabulary for citations; Dublin Core (a separate RDF vocabulary) is also catalogued there.
  • government-civictech — DCAT 3 is dual-listed there as the government open-data catalogue vocabulary and here as an RDF-stack vocabulary.
  • api-description — RDF Shapes and JSON Schema overlap as constraint languages over data; SHACL plays the equivalent role for RDF that JSON Schema plays for JSON.
  • healthcare-clinical — HL7 FHIR ships an RDF/Turtle profile alongside its native JSON/XML, and SNOMED-CT is increasingly published with OWL.
  • learning-content — Schema.org Educational profile, LRMI, and IMS Global Caliper use the JSON-LD stack.
  • graph-log-event-query — SPARQL is the RDF-graph cousin of Cypher/Gremlin/GQL; the family-overview there discusses the RDF-vs-LPG cultural split.
  • geospatial — GeoSPARQL is cross-listed; the W3C/OGC SSN/SOSA ontologies for sensors also live in both.

Tier 3 family table — RDF serialisations

FormatFirst appearedOriginTypeStatus (2026)URL
RDF 1.2 (abstract model)2025 (CR target, Rec NET 2026-05)W3C RDF & SPARQL WGAbstract data modelW3C Candidate Recommendation; not Rec before May 2026; merges RDF-star (triple terms) + directional language-tagged strings into corehttps://www.w3.org/TR/rdf12-concepts/
RDF 1.1 (predecessor)2014W3C RDF WGAbstract data modelActive Recommendation; will be superseded by 1.2https://www.w3.org/TR/rdf11-concepts/
RDF/XML1999 (rec.), 2004 revised, 2014 revisedW3CXML serialisation of RDFLegacy but still W3C-supported; widely deprecated in favour of Turtle/JSON-LD; remains the default for some OWL toolchainshttps://www.w3.org/TR/rdf-syntax-grammar/
Turtle (TTL) 1.2Turtle 1.1 in 2014; Turtle 1.2 alongside RDF 1.2 (2025 WD)W3C; orig. Dave Beckett & Tim Berners-LeeText serialisation of RDFDominant text serialisation; 1.2 WD adds triple-term and directional-language-tagged-string syntaxhttps://www.w3.org/TR/rdf12-turtle/
TriG 1.2TriG 1.1 in 2014; TriG 1.2 (2025 WD)W3CTurtle extended for named graphs (quads)Active; the multi-graph cousin of Turtlehttps://www.w3.org/TR/rdf12-trig/
N-Triples 1.2N-Triples 1.1 in 2014; 1.2 (2025 WD)W3CLine-based simple RDFActive; the “lowest common denominator” for streaming and toolinghttps://www.w3.org/TR/rdf12-n-triples/
N-Quads 1.2N-Quads 1.1 in 2014; 1.2 (2025 WD)W3CLine-based RDF with graph nameActive; ideal for dataset dumps and pipelineshttps://www.w3.org/TR/rdf12-n-quads/
JSON-LD 1.11.0 in 2014; 1.1 W3C Rec in 2020W3C JSON-LD WG; orig. Manu Sporny, Markus Lanthaler, Dave LongleyJSON-based linked-data serialisationVery active; dominant on the web via Schema.org; works alongside RDF 1.2https://www.w3.org/TR/json-ld11/
JSON-LD 1.1 Processing Algorithms (Expansion, Compaction, Flattening, Framing)2020 (Rec)W3CAlgorithmic DSL for JSON-LD transformationsActive Rec; framing in particular is a small declarative DSL for shaping JSON outputhttps://www.w3.org/TR/json-ld11-api/
RDFa 1.11.0 in 2008; 1.1 in 2012 (HTML5-aligned)W3CRDF embedded as attributes in HTML/XHTMLLegacy / niche; largely supplanted by JSON-LD-in-<script> for SEO but still W3C Rechttps://www.w3.org/TR/rdfa-primer/
RDF-star / SPARQL-star (community)2014 (Hartig), 2021 W3C CG Final ReportOlaf Hartig; W3C RDF-star CGRDF + SPARQL extension for statements-about-statementsBeing absorbed into RDF 1.2 / SPARQL 1.2; the CG report remains the canonical citation for the pre-1.2 erahttps://www.w3.org/2021/12/rdf-star.html

Tier 3 family table — Ontology languages

FormatFirst appearedOriginTypeStatus (2026)URL
RDF Schema (RDFS)1999; 1.2 alongside RDF 1.2 (2025 WD)W3CLightweight schema vocabulary (classes, properties, subClassOf)Active; the foundation under OWL; many practical KGs use RDFS+SKOS onlyhttps://www.w3.org/TR/rdf12-schema/
OWL 22009 first rec, 2012 OWL 2 second editionW3C OWL WGWeb Ontology Language (Description Logic family)Active Recommendation; no OWL 3 in active development; remains the formal ontology standardhttps://www.w3.org/TR/owl2-overview/
OWL 2 DL2009/2012W3COWL 2’s full Description Logic profile (SROIQ(D))Active; the reference profile; reasoners include HermiT, Pellet, FaCT++, ELKhttps://www.w3.org/TR/owl2-overview/
OWL 2 EL2009/2012W3CProfile for tractable subsumption — SNOMED-CT scaleHeavily used in biomedicine (SNOMED-CT, NCI Thesaurus)https://www.w3.org/TR/owl2-profiles/#OWL_2_EL
OWL 2 QL2009/2012W3CProfile for query rewriting over relational DBsActive, niche; used in OBDA (Ontology-Based Data Access) systems like Ontophttps://www.w3.org/TR/owl2-profiles/#OWL_2_QL
OWL 2 RL2009/2012W3CProfile expressible as forward-chaining rulesActive; the “production rules” profile, often run on Jena or RDFoxhttps://www.w3.org/TR/owl2-profiles/#OWL_2_RL
OWL 2 Manchester syntax2009/2012W3C / Univ. of ManchesterHuman-readable OWL syntaxActive; the default Protégé display formhttps://www.w3.org/TR/owl2-manchester-syntax/
OWL 2 Functional syntax2009/2012W3CFormal mathematical-style OWL syntaxActive; canonical normative formhttps://www.w3.org/TR/owl2-syntax/
SKOS2009 (Rec)W3C SWD WGSimple Knowledge Organization System — thesauri, taxonomies, controlled vocabsVery active; the de facto vocabulary for library/archive controlled vocabshttps://www.w3.org/TR/skos-reference/

Tier 3 family table — Shape constraints / validation

FormatFirst appearedOriginTypeStatus (2026)URL
SHACL 1.0 Core2017 (Rec)W3C Data Shapes WGConstraint validation language over RDF graphsActive Recommendation; the W3C-standard shape languagehttps://www.w3.org/TR/shacl/
SHACL 1.0 SPARQL Extensions2017 (Rec)W3CSPARQL-based custom constraints in SHACLActive Rechttps://www.w3.org/TR/shacl/
SHACL Advanced Features (SHACL-AF)2017 (Note)W3CFunctions, rules, node expressions on top of SHACLW3C Note; widely implemented in practice; being formalised as SHACL 1.2 Rules + Node Expressionshttps://www.w3.org/TR/shacl-af/
SHACL 1.2 Core2025 (FPWD)W3C Data Shapes WGSuccessor to SHACL 1.0; aligns with RDF 1.2W3C First Public Working Draft (March 2025)https://www.w3.org/TR/shacl12-core/
SHACL 1.2 SPARQL Extensions2025 (FPWD)W3CSuccessor to SHACL 1.0 SPARQL ExtensionsW3C FPWDhttps://www.w3.org/TR/shacl12-sparql/
SHACL 1.2 Rules2025 (FPWD, Dec 2025)W3CInference rules in SHACL — derive triples from shapesW3C FPWDhttps://www.w3.org/TR/shacl12-rules/
SHACL 1.2 Node Expressions2026 (FPWD)W3CFirst-class node-expression language for SHACLW3C FPWD (early 2026)https://w3c.github.io/data-shapes/shacl12-node-expressions/
ShEx 2.1 (Shape Expressions)1.0 in 2014; 2.0 Candidate 2017; 2.1 Final CG Report Oct 2019W3C ShEx Community GroupAlternative shape-constraint language; not W3C Rec trackActive CG, lower velocity than SHACL; still favoured in Wikidata/biomedical communities (Wikidata EntitySchemas use ShEx)https://shex.io/shex-semantics/

Tier 3 family table — Mapping / transformation

FormatFirst appearedOriginTypeStatus (2026)URL
R2RML2012 (Rec)W3C RDB2RDF WGRelational-to-RDF mapping languageActive W3C Rec; widely supported by virtual-RDF tools (Ontop, Morph-RDB, Stardog)https://www.w3.org/TR/r2rml/
RML 1.0 (RDF Mapping Language)First publication 2014; RML 1.0 spec under Knowledge Graph Construction W3C CG, 2024 draftKG Construction W3C Community Group (Ghent/IDLab, Anastasia Dimou et al.)Superset of R2RML; maps CSV/JSON/XML/databases into RDFActive; the de facto modern mapping language; portal at w3id.org/rml/https://w3id.org/rml/portal/
RML-star2021–2024KG Construction W3C CGRML extension for RDF-star outputActive, aligns with RDF 1.2’s triple termshttps://w3id.org/rml/star/spec
YARRRML2018imec/Ghent (Pieter Heyvaert, Ben De Meester)YAML-based human-friendly serialisation of RML/R2RMLActive; the practical authoring layer; YARRRML-star extension for RDF-starhttps://rml.io/yarrrml/
SPARQL-Generate / SPARQL-Anything2016 / 2021Maxime Lefrançois (Mines Saint-Étienne) / SPARQL Anything projectSPARQL-based generation of RDF from any data sourceActive, alternative to RMLhttps://ci.mines-stetienne.fr/sparql-generate/
RDB2RDF Direct Mapping2012 (Rec)W3C RDB2RDF WG”Default” mapping from any relational schema to RDFActive Rec, used as the fallback when no R2RML is writtenhttps://www.w3.org/TR/rdb-direct-mapping/

Tier 3 family table — Query / SPARQL family

FormatFirst appearedOriginTypeStatus (2026)URL
SPARQL 1.1 Query2013 (Rec)W3CQuery language for RDFActive Recommendation; deployed everywherehttps://www.w3.org/TR/sparql11-query/
SPARQL 1.1 Update2013 (Rec)W3CWrite/update operations on RDF datasetsActive Rechttps://www.w3.org/TR/sparql11-update/
SPARQL 1.1 Federated Query2013 (Rec)W3CSERVICE clause for cross-endpoint queriesActive Rechttps://www.w3.org/TR/sparql11-federated-query/
SPARQL 1.1 Protocol2013 (Rec)W3CHTTP protocol for SPARQL endpointsActive Rechttps://www.w3.org/TR/sparql11-protocol/
SPARQL 1.1 Graph Store Protocol2013 (Rec)W3CHTTP CRUD on named graphsActive Rec; 1.2 WD published Dec 2024https://www.w3.org/TR/sparql11-http-rdf-update/
SPARQL 1.2 Query2025-08 WDW3C RDF & SPARQL WGSuccessor to SPARQL 1.1; integrates SPARQL-star (triple terms), better aggregation, JSON-LD result fmtW3C Working Draft; tracks RDF 1.2 timeline (Rec NET 2026-05)https://www.w3.org/TR/sparql12-query/
SPARQL 1.2 Update2025 WDW3CUpdate for SPARQL 1.2W3C WDhttps://www.w3.org/TR/sparql12-update/
SPARQL 1.2 Federated Query2026-01 WDW3CFederation for SPARQL 1.2W3C WDhttps://www.w3.org/TR/sparql12-federated-query/
GeoSPARQL 1.1OGC 22-047, published 2024Open Geospatial ConsortiumSpatial extension to SPARQL + spatial ontologyActive OGC Standard; adds spatial measurement properties, FeatureCollection, SHACL validation, W3C Profile designationhttps://docs.ogc.org/is/22-047r1/22-047r1.html
GeoSPARQL 1.02012OGCOriginal spatial extension to SPARQLSuperseded by 1.1, still implemented in older systemshttps://www.ogc.org/standards/geosparql/

Tier 3 family table — Vocabularies / linked-data services

FormatFirst appearedOriginTypeStatus (2026)URL
Schema.org2011 launch; v30.0 released 2026-03-19Google, Bing, Yahoo, Yandex (W3C Schema.org CG)Cross-domain vocabulary, JSON-LD-firstVery active; the most-deployed RDF vocabulary on the web (driven by SEO/structured-data)https://schema.org/
Dublin Core / DCMI Metadata Terms1995 (informal); current 2020 DCMIDCMIGeneral-purpose bibliographic/metadata vocabularyActive; foundational vocab for libraries, repositories, DCAT, etc.https://www.dublincore.org/specifications/dublin-core/dcmi-terms/
VOID (Vocabulary of Interlinked Datasets)2011 (W3C Note)W3C SWIGRDF vocabulary for describing RDF datasets themselvesActive Note; widely used in LOD cloud metadatahttps://www.w3.org/TR/void/
DCAT 32014 (DCAT 1), 2020 (DCAT 2), 2024 DCAT 3 W3C RecW3C DXWGDataset catalogue vocabulary (government open-data driver)Active Rec; also catalogued in government-civictechhttps://www.w3.org/TR/vocab-dcat-3/
PROV-O (ontology)2013 (Rec)W3C Provenance WGOWL ontology for provenanceActive Rechttps://www.w3.org/TR/prov-o/
PROV-N2013 (Rec)W3CHuman-readable notation for PROVActive Rechttps://www.w3.org/TR/prov-n/
PROV-DM2013 (Rec)W3CAbstract data model for provenanceActive Rechttps://www.w3.org/TR/prov-dm/
Linked Data Notifications (LDN)2017 (Rec)W3C Social Web WGHTTP protocol for delivering RDF notificationsActive Rec; underpins ActivityPub-adjacent and Solid notificationshttps://www.w3.org/TR/ldn/
Solid Protocol2018+; v0.12.0 ED dated 2025-04-12Tim Berners-Lee, MIT/Inrupt, Solid Community GroupDecentralised personal-data pods over RDF + LDP + WebID + OIDCActive CG draft; Inrupt commercial implementation; not yet W3C Rechttps://solidproject.org/TR/protocol
SSN / SOSA2017 (Rec)W3C/OGCSemantic Sensor Network ontologyActive Rec; used in IoT and earth-observation KGshttps://www.w3.org/TR/vocab-ssn/

Notable threads

  • RDF 1.2 and SPARQL 1.2 — the long-promised triple-term unification. The single biggest pain point in classical RDF was reification: how do you say “Alice claims Bob is a friend of Carol since 2020” without inventing a synthetic node for the statement itself? Reification (rdf:Statement), n-ary relations, named graphs, and singleton properties were all partial answers; none felt native. Olaf Hartig’s 2014 RDF-star proposal — allow a triple as a term inside another triple — got a 2021 W3C Community Group Final Report and is now being absorbed into RDF 1.2 and SPARQL 1.2 (both at Working Draft / Candidate Recommendation in 2025–2026, Rec target no earlier than May 2026). This is the most significant change to the core RDF data model since 1999. Turtle 1.2, TriG 1.2, N-Triples 1.2, N-Quads 1.2 are all WD revisions that add triple-term syntax in lockstep.

  • Turtle’s quiet victory as the canonical text serialisation. When RDF 1.0 shipped in 1999 with only RDF/XML, it was nearly unusable by hand — rdf:about attributes nested inside rdf:Description elements made even small ontologies illegible. Turtle (Dave Beckett, 2004) eventually won the readability war and was elevated to a W3C Recommendation alongside RDF 1.1 in 2014. Today every W3C RDF document, every OWL ontology, every SHACL shape, every Wikidata Lexeme dump, and the bulk of teaching materials use Turtle by default. RDF/XML survives mainly because some legacy OWL tooling still emits it.

  • OWL 2 Profiles as a successful pragmatic split. OWL 1 (2004) made the classic Description-Logic mistake of offering one “full” language whose decidability properties varied by which constructs you used. OWL 2 (2009, second edition 2012) responded with profiles — DL (the full reference profile, decidable in N2EXPTIME), EL (subsumption in polynomial time, the SNOMED-CT-scale profile), QL (designed for query rewriting over relational DBs), and RL (expressible as forward-chaining production rules). This is a model worth studying — instead of one language with murky complexity classes, four named profiles with crisp tractability guarantees and dedicated reasoners (ELK for EL, Ontop for QL, RDFox for RL, HermiT/Pellet for DL). The profile approach is why OWL is still in production at biomedical and pharma scale fifteen years after publication.

  • SHACL vs ShEx — SHACL won at W3C, but ShEx is far from dead. Both shape languages emerged around 2014–2015 to answer the same question: how do you validate that an RDF graph has the structure your application expects? SHACL went the W3C Working Group route (2017 Rec), won institutional buy-in, and is now seeing SHACL 1.2 First Public Working Drafts (Core, SPARQL Extensions, Rules, Node Expressions, 2025–2026). ShEx stayed in W3C Community Group land (2.1 Final CG Report October 2019) and never became a Recommendation, but it remains the validation language of choice for Wikidata EntitySchemas and for several biomedical KG projects, and its compact syntax is widely considered more authorable than SHACL’s. The two communities now broadly co-exist; tooling like Apache Jena supports both.

  • RML and YARRRML — the modern “lift anything into RDF” pipeline. R2RML (W3C Rec 2012) only handled relational databases. RML, developed at IDLab Ghent and now under the W3C Knowledge Graph Construction Community Group (RML 1.0 spec under active development through 2024, with a stable portal at w3id.org/rml/), generalises R2RML to any structured source: CSV, JSON, XML, HTML tables, even REST APIs via logical-source extensions. YARRRML (2018) is a YAML-flavoured surface syntax for the same model — RML written in Turtle is dense; YARRRML written in YAML is editable by data engineers. The combination has become the practical answer to “we have a thousand heterogeneous data sources, lift them into one knowledge graph” — far more so than per-source bespoke scripts.

  • Schema.org’s quiet dominance via JSON-LD in HTML. Google, Bing, Yahoo, and Yandex launched Schema.org in 2011 as a single shared vocabulary for marking up web pages so search engines could extract structured data. Sixteen years and 30 major releases (v30.0, 2026-03-19) later, Schema.org is by a wide margin the most deployed RDF vocabulary on the planet — embedded as <script type="application/ld+json"> blocks in tens of millions of pages — and most of its users have no idea they’re using RDF. This is the Semantic Web’s commercial victory hiding in plain sight: the format that won was the one that piggybacked on SEO incentives rather than on ontologists’ aesthetic preferences.

  • Solid Protocol and the personal-data revival. Tim Berners-Lee has spent the post-2015 period trying to reposition the RDF stack as the substrate for personal-data sovereignty — individual users hosting Pods (Linked Data Platforms) and authorising third-party apps to read/write specific graphs. Solid Protocol 0.12 (Editor’s Draft April 2025) layers WebID-OIDC authentication, Linked Data Platform CRUD, and access-control rules (WAC, then ACP) onto Turtle and JSON-LD. Inrupt (Berners-Lee’s company) drives the commercial side; the EU-funded NGI ecosystem and several national-ID pilots (Flemish Solid-OS, French Data4Citizens) are the public-sector showcases. Whether Solid succeeds remains open, but it represents the strongest current attempt to make decentralised data on the open web a default rather than a curiosity.

  • The knowledge-graph era as semantic-web vindication. When Google formally announced “things, not strings” with the Knowledge Graph in 2012, it validated the entire 1999–2010 RDF research programme without saying the word “RDF” once. Wikidata (2012+) became the largest open RDF dataset in existence. Enterprise KG platforms — Stardog, GraphDB, Neptune (Amazon), Anzo, TopBraid, RDFox — built sustainable businesses on SPARQL + OWL + SHACL. Bloomberg, eBay, Springer Nature, AstraZeneca, Bayer, and most major pharma all run production KGs. The LLM-RAG wave (2023+) added a fresh round of demand: KGs as grounded retrieval substrates feeding LLM context windows. The Semantic Web didn’t fail — it just took twenty years to ship and now ships under different marketing.

Citations

Graph View