feat: auto-generate RDF/XML and JSON-LD serialisations from TTL source

[Daham-Mustaf]

Context

The mobilityDCAT-AP ontology is currently maintained in three separate files:

• mobilitydcat-ap_v1.1.0.ttl — Turtle (hand-edited)
• mobilitydcat-ap_v1.1.0.rdf — RDF/XML (hand-edited)
• mobilitydcat-ap_v1.1.0.jsonld — JSON-LD (hand-edited)

All three are edited independently, which leads to inconsistencies — the bugs
found in issue #179 exist because the three files have drifted from each other.

Proposal

Single source of truth: the TTL file.

The RDF/XML and JSON-LD serialisations should be automatically generated from
the TTL file using a script. This means:

• Maintainers only edit mobilitydcat-ap_v1.1.0.ttl
• A script converts it to .rdf and .jsonld
• The generated files are committed to the repository
• CI can verify the generated files are up to date with the TTL source

Implementation

A script scripts/serialise_ontology.py using rdflib:

from rdflib import Graph

g = Graph()
g.parse("mobilitydcat-ap_v1.1.0.ttl", format="turtle")
g.serialize("mobilitydcat-ap_v1.1.0.rdf", format="xml")
g.serialize("mobilitydcat-ap_v1.1.0.jsonld", format="json-ld")

Benefits

• Single point of edit — no more three-way inconsistencies
• Bugs fixed in TTL automatically propagate to all formats
• Simpler review — only TTL needs to be reviewed in PRs
• Can be integrated into CI to auto-regenerate on TTL change

Relation to existing scripts

The repository already has scripts/serialise-vocabulary.sh and
scripts/serialise-examples.sh. This would add a dedicated ontology
serialisation script following the same pattern.

Related

• #179 — bugs caused by manual three-file maintenance
• #180 — PR fixing those bugs across all three files manually (last time)