Construct-DCAT

Construct-DCAT is a lightweight RDF vocabulary for adding representation-independent semantic anchors to DCAT catalog records for heterogeneous construction datasets. The catalog metadata is RDF, but the data it describes may be IFC, Asset Administration Shell (AAS) JSON, tabular, geospatial, sensor, document, RDF, or another representation.

Identity and status

• Vocabulary IRI: https://w3id.org/construct-dcat
• Namespace: https://w3id.org/construct-dcat#
• Preferred prefix: cx:
• Version: 0.1.0
• Status: initial conceptual release

Documentation

The full, navigable vocabulary specification is published at jundahuang9123.github.io/ConstructDCAT. Stable publication paths are also available for the 0.1.0 release and the latest release.

The authoring source is spec/index.html. ReSpec generates a frozen static snapshot, so visitors do not need JavaScript or a live ReSpec dependency. To rebuild and synchronize the root, latest, and versioned copies, use Node.js 24 or later and pnpm 11.9.0:

pnpm install
pnpm docs:build

Scope

Construct-DCAT extends descriptions of dcat:Dataset records with a small set of construction-oriented discovery anchors. It can be used in DCAT and DCAT-AP catalogs, but version 0.1.0 is not a complete DCAT-AP profile and defines no SHACL constraints. cx:AASDataset and cx:BIMDataset are initial examples, not an exhaustive dataset taxonomy.

The four properties do not hard-code a closed list of external vocabularies. Their objects are IRIs from any suitable ontology, class hierarchy, AAS model, or controlled vocabulary. This release asserts no rdfs:subPropertyOf alignments to DCAT or Dublin Core.

Current terms

Term	Kind	Domain or parent	Range	Meaning
cx:AASDataset	Class	subclass of dcat:Dataset	—	A dataset represented by, or organized around, one or more Asset Administration Shells.
cx:BIMDataset	Class	subclass of dcat:Dataset	—	A dataset containing building information modeling data.
cx:describesAssetType	Object property	dcat:Dataset	rdfs:Class	The physical asset or building-element type described by the dataset.
cx:usesOntology	Object property	dcat:Dataset	no formal range	An ontology, vocabulary, or semantic resource used or referenced by the dataset. Values may identify OWL ontologies, SKOS concept schemes, or other resolvable resources.
cx:alignsWithConcept	Object property	dcat:Dataset	skos:Concept	A controlled concept used to classify or semantically align the dataset.
cx:hasAASSubmodel	Object property	dcat:Dataset	no formal range	An AAS submodel or submodel descriptor exposed as a catalog-level discovery anchor. A range is deferred until a stable canonical RDF class IRI is selected.

Turtle example

@prefix bot:  <https://w3id.org/bot#> .
@prefix cx:   <https://w3id.org/construct-dcat#> .
@prefix dcat: <http://www.w3.org/ns/dcat#> .
@prefix ex:   <https://example.org/construct-dcat/> .
@prefix ifc:  <https://standards.buildingsmart.org/IFC/DEV/IFC4_3/OWL#> .

ex:wall-model
    a dcat:Dataset, cx:BIMDataset ;
    cx:describesAssetType ifc:IfcWall ;
    cx:usesOntology bot: ;
    dcat:distribution ex:wall-model-ifc .

This RDF describes the catalog record. Its distribution can remain a non-RDF IFC-SPF file; the same principle applies to AAS JSON, CSV, geospatial files, sensor exports, and documents. See the complete example.

Construct-DCAT and VoID

VoID can describe vocabularies, classes, properties, and partitions occurring within RDF datasets. Construct-DCAT instead exposes representation-independent semantic anchors on DCAT catalog records, including records for non-RDF datasets. RDF datasets may use both vocabularies together.

Repository layout

ConstructDCAT/
├── .github/workflows/validate.yml
├── docs/
│   ├── construct-dcat.jsonld
│   ├── construct-dcat.ttl
│   └── index.html
├── examples/example-catalog.ttl
├── queries/wall-bot.rq
├── scripts/validate.py
├── .gitignore
├── CITATION.cff
├── LICENSE
├── README.md
├── construct-dcat.jsonld
├── construct-dcat.ttl
└── requirements.txt

Validation

python -m pip install -r requirements.txt
python scripts/validate.py

The validator parses both vocabulary serializations and every GitHub Pages copy, checks the six-term model, confirms the serializations are RDF-isomorphic, verifies the static documentation snapshots and stable term fragments, and parses and executes the SPARQL query. The copies under docs/ let the W3ID content-negotiation targets return RDF-specific media types; the canonical editable RDF sources remain the repository-root files.

License and maintainer

Construct-DCAT is licensed under Creative Commons Attribution 4.0 International.

Maintained by Junda Huang, University of Wuppertal. Contact: huang@uni-wuppertal.de · GitHub: jundahuang9123