OPDA ontology-model documentation — information architecture
This directory specifies how OPDA's ontology model is documented. The ontology model itself (the 24 emitted TTL files at source/03-standards/ontology/, the 23 SKOS schemes, the 40 minted classes, the BASPI5 profile) lives elsewhere; the artefacts under this directory specify the structure of the docs that describe it.
Four documentation outputs
OPDA's ontology model is documented as four tiers. Each tier has its own audience, its own conventions, and its own IA spec in this directory.
| Tier | Audience | What it shows | IA spec |
|---|---|---|---|
| Concept | Property-industry SMEs (surveyors, conveyancers, lenders, government data leads) | What each business object means and why; non-technical narrative + diagrams | concept-model-ia.md |
| Logical | Data engineers, solution architects, integrators | Entity-relationship view: typed attributes + cardinalities + relationships, platform-independent | logical-model-ia.md |
| Physical — deployment / database | Triplestore operators, SPARQL consumers, ontology integrators, devops | Named-graph layout, derived consumer profiles, content-negotiation, BASPI5 overlay composition, CI gates — the deployed/served form of the ontology at opda.org.uk/pdtf/ |
physical-database-ia.md |
| Physical — ontology | Ontology engineers, SHACL implementers, SPARQL consumers, regulators | OWL/SHACL/SKOS Turtle layout: per-module classes, three-graph separation, severity tiers, SHACL-AF rules, overlay profiles | physical-ontology-ia.md |
The four tiers are linked but independent. A reader of the Concept tier never needs to read Turtle; a reader of the Physical-Ontology tier should be able to trace any class back to the Concept-tier narrative via dct:source URIs.
Why four tiers
OPDA's stakeholders span business and engineering audiences with overlapping but distinct needs:
- Concept tier answers "what does an OPDA Property mean, and why does Identity-Criterion (IC) matter?" — narrative-first, no jargon.
- Logical tier answers "what attributes does a Property have, and how does it relate to Address, LegalEstate, Transaction?" — schema-shape without commitment to RDF, JSON-LD, or any particular serialisation.
- Physical-DB tier answers "which named graph holds
opda:Property, what derived consumer profiles include it, and how do I fetch it fromopda.org.uk/pdtf/Propertyvia content negotiation?" — deployment + serving navigation. - Physical-Ontology tier answers "which Turtle file emits
opda:Property, what SHACL shapes constrain it, and whatdct:sourceURIs trace its definition?" — RDF / SHACL / SKOS specifics in source-tree form.
Skipping a tier produces predictable friction: SMEs reading Turtle, engineers reading marketing prose, devops guessing at named-graph layout from OWL declarations. The four tiers de-couple those audiences without losing traceability.
Cross-tier traceability requirement
Every entity in every tier MUST carry a stable URI that resolves across all four tiers. The Physical-Ontology tier uses Turtle's dct:source URI; the Concept / Logical / Physical-DB tiers carry the equivalent URI in their own format conventions (per the per-tier IA specs).
Concretely: anyone reading the Concept-tier description of "Property" should be able to follow a URI to the Logical-tier entity row, then to the Physical-DB JSON path, then to the Physical-Ontology TTL block — without name collisions, without orphaned references.
The four IA specs each define how that URI surfaces in their tier.
Source of truth for documentation generation
Every tier generates from the emitted ontology at source/03-standards/ontology/ (the 24 TTLs + the 15 paired exemplars + the BASPI5 profile). The A9 per-kind discipline (per ADR-0007 §"A9 per-kind discipline output") was designed precisely so the TTL is self-sufficient: every owl:Class carries rdfs:comment (IC + hard cases verbatim) + skos:scopeNote (UFO/DOLCE citation) + dct:source (audit-trail pointer).
Two things explicitly not documentation sources:
- The ODR corpus (
docs/ontology/odr/) — it is the Council deliberation audit trail. The ontology embeds the outcome via A9; the ODR explains why the Council ratified that outcome. Documentation generation usesdct:sourceURIs as link targets only, never as content sources. - The PDTF JSON Schemas (
source/03-standards/schemas/) — they were upstream input to the Council programme, not output to be documented in the 4-tier scheme. They have their own documentation in the nested schemas repo.
Governance
This IA spec governs the documentation of OPDA's ontology model, not the ontology model itself. The model is governed by the ontology-implementation programme:
- ADR-0007 — Ontology generator specification
- ADR-0008 through ADR-0014 — the 7-ADR implementation programme that emitted the current ontology
- ODR-0001 — Linked Data Council methodology — governs modelling decisions
- ODR-0011 — Enumeration vocabularies — 7-category UFO meta-vocabulary that the Concept + Logical tiers must surface
When this IA spec lands as accepted, the per-tier docs themselves become the next deliverable: the IA tells how the generator should emit them; the generator extension is a small docs-gen mode on top of opda-gen, not a separate authoring workstream.
File inventory
docs/information-architecture/
├── README.md This file
├── concept-model-ia.md Concept-tier IA spec
├── logical-model-ia.md Logical-tier IA spec
├── physical-database-ia.md Physical (JSON/database) tier IA spec
└── physical-ontology-ia.md Physical (ontology) tier IA spec
Provenance
Authored 2026-05-28 in OPDA's namespace and Council idiom. The four-tier pattern is a common documentation convention in industrial data-modelling practice; the IA specs here are OPDA-native and reference only OPDA's accepted ADRs / ratified ODRs.
Comments