Foundational-ontology choice: UFO-as-lens, scoped to the Relator spine

Context and Problem Statement

The OPDA PDTF ontology rests, in part, on Giancarlo Guizzardi’s Unified Foundational Ontology (UFO): the seven-category framework is carried as advisory opda:ufoCategory SKOS tags (ODR-0011 §8a), a gated set of gUFO rdf:type gufo:Quality markers types five descriptive leaves (ADR-0034), and OntoClean’s rigidity / identity / dependence tests are the decision procedure for “OWL subclass vs coded SKOS facet” (ODR-0027). A four-stream sourced review (published critiques of UFO/OntoUML/gUFO/OntoClean; BFO and the UFO-vs-BFO debate; DOLCE/SUMO/GFO/gist and the lightweight alternatives; and the meta-question of whether a data-interchange standard should rest on a foundational ontology at all), published as the /ontology/foundational-ontology reference page, surfaced a real decision: should the standard retain UFO, deepen it, switch to BFO, or drop the foundational layer entirely?

The decision is not whether to import a heavyweight foundation — OPDA never has. UFO is used as a restrained design lens: reasoning is shallow (only rdfs:subClassOf type-propagation fires; ODR-0029), domain/range are validated as SHACL rather than inferred, and the gUFO annotations are quarantined to the annotation graph (ODR-0010 §Q7a, ODR-0004 §3a), causally inert in the emitted artefact. Two facts sharpen the question: the layer is a UFO+DOLCE hybrid (35 of 47 ufoCategory tags read “Quale-in-Region”, a DOLCE construct), and the strongest published critique is the gUFO authors’ own — UFO’s modal/mereological apparatus cannot be captured in decidable OWL 2 DL, so constraints belong in SHACL. The live question is therefore: is a lens you barely materialise worth its conceptual and maintenance tax, and where exactly does it earn its keep?

Decision Drivers

• Domain fit. A conveyance is a lattice of roles, claims, obligations, evidence and legal estates; UFO-C (social commitments/claims) and UFO-L (Hohfeldian Claim-Right↔Duty, Power↔Subjection) model these as first-class relational endurants. DOLCE lacks a native relator; BFO must reify obligations as directive information entities.
• Decidability / soundness. The gUFO authors (Almeida, Guizzardi, Sales & Fonseca, 2026) show UFO is not faithfully expressible in OWL 2 DL — constraints must move to SHACL/SPARQL. This validates the existing ODR-0029 architecture and argues against deepening.
• No measured benefit, real complexity cost. Bernabé et al. (2023) found no measurable downstream benefit from foundational ontologies in practice; complexity is the universal cost — an argument against any heavyweight commitment.
• Revealed preference of real standards. No property/built-environment interchange standard (PDTF, RICS/OSCRE, IFC/EXPRESS — ifcOWL being a widely-criticised after-the-fact projection) rests on a foundational ontology; the successful interchange standards (schema.org, FHIR, GS1, ISO 20022) are pragmatic.
• Anti-overclaim doctrine. OPDA’s own S035/S036 + the ODR-0029 “RDFS-Plus” rename precedent forbid asserting more than the artefact performs.
• Standardisation gap. BFO’s only decisive advantage is ISO/IEC 21838-2 standing — relevant only to external interoperability credibility, not modelling correctness.

Considered Options

• Option A (chosen) — Retain UFO as a design-time lens, scoped to the Relator/Role/RoleMixin spine. Keep OntoClean as the subclass-vs-facet decision procedure and the gUFO/ufoCategory tags as quarantined, inert provenance; relabel the register-deference tags honestly; layer the standards alignment by maturity.
• Option B — Deepen to a heavyweight OntoUML / full-UFO import. Rejected: UFO’s apparatus is not capturable in decidable OWL 2 DL (gUFO authors, 2026), so deepening buys modelling ceremony and an undecidability problem with no payoff a property-transaction file can cash.
• Option C — Switch the foundation to BFO. Rejected as a switch: BFO has no native legal/deontic apparatus and must reify a Claim-Right as a directive information entity; its only decisive win (ISO/IEC 21838-2) is captured by the layered Q3 reserve below without abandoning UFO-L’s superior domain fit (and Merrill 2010 shows BFO’s interoperability benefits do not require its contested realism).
• Option D — Drop the foundational layer entirely (pragmatic SHACL+SKOS only). Rejected as the standing decision, but recorded as the Devil’s Advocate’s held-as-live position: OntoClean is separable from UFO and the vocabulary is droppable — but the Relator category is categorially load-bearing (OntoClean is a theory of meta-properties and is silent by construction on “the Relator founds the RoleMixins”), so the foundation is optional, not harmful, and is retained pending the re-open triggers.

Decision Outcome

Chosen option: “Retain UFO as a design-time lens, scoped to the Relator/Role/RoleMixin spine”, because that is the precise locus where UFO does categorial work that OntoClean + SHACL + SKOS cannot reconstruct — and it is exactly the relational-endurant layer where UFO-L’s Hohfeldian apparatus, the standard’s strongest domain-fit asset, also lives.

The council’s adjudication (session-040; Queen Allemang, DA Cagle + Guizzardi, Guarino, Knublauch, Smith) was unanimous after a live cross-examination, but it scoped the affirmation rather than rubber-stamping it. The honest account of how UFO is used: OPDA’s modelling process is governed by a UFO/OntoClean decision procedure; the artefact carries an inert provenance record of that governance. UFO is load-bearing only at the Relator/Role spine (≈ a handful of classes — opda:Proprietorship/opda:Transaction founding Seller/Buyer). The council’s co-signed refinement is sharper: the strictly irreducible thing is the relational-reification primitive (reify the relation as a first-class node that owns the relation’s aggregation/modal attributes — proven by the placement of opda:numberOfSellers on the Proprietorship Relator, opda-agent.ttl:155, which OntoClean’s monadic +R/+I/+D cannot derive, having no mediation primitive); UFO’s Relator is OPDA’s chosen, apt expression of that primitive, not the unique source (the corpus cites FIBO Arrangement as a co-precedent on opda:Transaction); and the bounded reason UFO earns the choice is the deontic core — only UFO+UFO-L natively unifies the reification primitive with Hohfeldian Claim-Right↔Duty correlativity, which this Hohfeld-throughout domain (charges, covenants, easements, offer/acceptance) needs, where FIBO supplies the shape but not the correlativity and BFO must reify a claim-right as a directive information entity. So the standing is necessary-and-bounded, not pervasive-and-grounding. OntoClean (separable from UFO, predating it) carries the rigidity/identity calls; and the 35 “Quale-in-Region” tags are register-deference (an external authority — EPC, HMLR, INSPIRE — closes the value space), not UFO doing categorial work, so they drop the UFO byline. The wire format reasons with nothing UFO-shaped, so the published claim is “UFO-informed”, never “UFO-grounded with guarantees.”

Consequences

• Good, because the standard keeps the one discipline that demonstrably moved emitted bytes (the OntoClean criterion adjudicated tenureKind, VouchEvidence, and RiskAssessment) while shipping only pragmatic SHACL/SKOS/JSON that stands on its own.
• Good, because it commits no contested metaphysics to the reasoner (the gUFO quarantine), so there is no “forked foundation” cost at runtime and the decision is reversible.
• Good, because the honesty dispositions (UFO-informed; the scopeNote lineage; UFO-C flagged least-mature) pre-empt the one credibility risk a working standard cannot afford — overclaiming a guarantee the architecture does not provide.
• Good, because the layered Q3 alignment captures BFO’s standardisation upside if and when it is actually required, without paying alignment-maintenance tax speculatively or abandoning UFO-L’s deontic fit.
• Bad (accepted), because the foundation layer can only be described honestly by enumerating its caveats — the DA’s standing point that this is “optional, not load-bearing except at the Relator spine” is recorded as live dissent with falsifiable re-open triggers.
• Neutral, because the retention is conditional: it stands only while the gUFO/ufoCategory annotations remain quarantined and CI-enforced outside the reasoned union — the moment that is breached, the retention lapses.

Confirmation

• The gUFO/ufoCategory quarantine stays CI-enforced: the three-graph separation gate (ASK { GRAPH opda:annotations { ?s a sh:NodeShape } } → FALSE, ODR-0004 §3a) and the ODR-0029 entailment-closure byte-identity gate both remain green; no gufo:/ufoCategory triple appears in a graph that drives inference or SHACL.
• The honesty dispositions are realised in the artefact: a machine-readable skos:scopeNote on opda:ufoCategory (the three-part disclosure below), the “UFO-informed” framing in the /ontology prose, and a UFO-C maturity flag — verifiable by re-emission under the byte-identity gate.
• The decision is proposed; the operator ratifies adoption. The council shapes the proposal, not its adoption.

Rules

The dispositions routed from session-040 (all status: proposed):

1. Scoped retention. UFO is retained as a design-time lens scoped to the Relator/Role/RoleMixin spine. The gUFO rdf:type markers and opda:ufoCategory tags remain annotation-graph-only (ODR-0010 §Q7a, ODR-0004 §3a); the ODR-0029 quarantine is load-bearing — retention lapses if it is breached.
2. Register-deference relabel. The ufoCategory tags whose value space is closed by an external authority (EPC bands, HMLR tenure/title codes, INSPIRE classes — the 35 “Quale-in-Region” majority) are coded sh:in by register-deference, not by a UFO/OntoClean judgement; documentation and the scopeNote must stop presenting them as UFO doing categorial work.
3. No deepening (Q2). Do not import OntoUML or full UFO; the gUFO authors’ OWL-2-DL undecidability finding is the technical warrant. Constraints stay in SHACL.
4. ADOPT-NOW — IAO information-artefact crosswalk (Q3). A thin, external, referenced-not-imported set of skos:exactMatch/closeMatch/rdfs:subClassOf edges from OPDA’s document/record family (title registers, EPCs, searches, requisitions, the AttachedDocument line) to obo:IAO (document / information-content-entity / directive), in the annotation graph, on the opda:EPCCertificate rdfs:subClassOf prov:Entity precedent — never owl:imports, never reasoned over.
5. HOLD-IN-RESERVE — BFO upper spine + ISO-21838 credential (Q3). Co-located with the IAO crosswalk (rule 4) in a single quarantined annotation-graph file — one file, two maturity clocks, the same prov:Entity reference-not-import mechanism. Write the crosswalk spec now; instantiate the edges and bank the “ISO/IEC 21838-aligned” credential only when a conformance clause exists on an actual contract/tender the bridge satisfies. Interim honest claim is the narrow true one (“OPDA carries deliberate IAO crosswalks”), never the broad inert one.
6. Deontic core stays UFO-L (Q3). Hohfeldian Claim-Right↔Duty (charges, covenants, easements, requisitions) remains UFO-L-shaped; the bridge never touches it.
7. Honesty dispositions (Q4). (a) Publish “UFO-informed”, not “UFO-grounded with guarantees”. (b) Add a skos:scopeNote on opda:ufoCategory stating: (i) the categories are UFO’s; (ii) UFO’s quality categories descend from DOLCE’s Quality / Quale / Region (Masolo et al., WonderWeb D18, 2003); (iii) the majority of OPDA’s tagged properties fall under that DOLCE-derived quality apparatus. (c) Flag UFO-C (social/intentional/legal) as the least-mature layer wherever the standard leans on it.
8. Anti-patterns (do NOT). Move the gUFO/ufoCategory annotations into a reasoned graph; import OntoUML or BFO; bank an “ISO-aligned” credential off inert edges; or present the register-deference quality tags as UFO doing categorial work.

Vote and Dissent

Council session-040 (Full Council; agent-fan-out with live SendMessage cross-talk). Per-question tally, all 6-0-0:

Held-as-live dissent (Cagle, DA). OntoClean is strictly separable from UFO, so the foundation vocabulary is droppable — the layer is optional, not harmful. He WITHDREW the REJECT push because net harm could not be shown at ~zero runtime cost; the dissent is recorded as live with three falsifiable re-open triggers: (i) the UFO/ufoCategory layer is ever made reasoned-over (moved out of the annotation-graph quarantine into a graph that drives inference or SHACL); (ii) the 35/47 DOLCE-under-a-UFO-label mismatch is shown to have caused a downstream modelling error or maintainer miscategorisation; (iii) maintaining the foundation mapping measurably delays or blocks a council decision. If any trigger fires, Option D (retire the UFO vocabulary; keep OntoClean-as-judgement + SHACL + SKOS) is revisited.

Knublauch’s standing condition. The AFFIRM is conditional on the gUFO quarantine remaining CI-enforced (three-graph separation + the ODR-0029 regime that excludes domain/range and cannot propagate the advisory types).

Amendments

2026-06-15 — Validation pass (Council session-044)

A validation council (session-044; Queen Allemang, DA Cagle + Baker, Guizzardi, Knublauch; agent-fan-out, with the convener web-verifying the bibliography as the decisive check) re-audited this record and session-040 for soundness, completeness, internal consistency, and citation integrity. The decision is unchanged — the keystone reasoning (UFO scoped to the Relator/Role spine; the relational-reification primitive; OntoClean separable; register-deference for the quality codes; “UFO-informed, not UFO-grounded”) was re-confirmed sound and corpus-grounded at file:line (numberOfSellers → opda:Proprietorship; the opda:Relator comment + the opda:Transaction FIBO-Arrangement co-precedent). Corrections recorded:

1. Date corrected 2026-06-14 → 2026-06-15. This record was committed 2026-06-15 (alongside ODR-0031 and sessions 041–042) and its own §More-Information cites a breach discovered 2026-06-15; the 06-14 stamp was an error inconsistent with the arc and with the record’s own body.
2. Citation re-verification — the gUFO reference is REAL (the council was wrong to suspect it). The validation personas, training-cutoff-limited, unanimously suspected “Almeida, Guizzardi, Sales & Fonseca, gUFO: A Gentle Foundational Ontology for Semantic Web Knowledge Graphs (2026), arXiv:2603.20948” was fabricated. Web verification on 2026-06-15 refuted this: the paper is genuine (arXiv:2603.20948, submitted 2026-03-21; indexed at arXiv + Univ. Twente) — a distinct, later work from the 2019 gUFO: A Lightweight Implementation of UFO (purl.org/nemo/gufo) that ADR-0034 cites. No deletion — the reference stands. All other external citations were web-confirmed real (Merrill, Applied Ontology 5(2), 2010; Bernabé et al., J. Biomedical Semantics 14:21, 2023, doi 10.1186/s13326-023-00300-z; Barcelos et al., ONTOBRAS 2013, CEUR Vol-1041; Masolo et al., WonderWeb D18, 2003). Two page-only paraphrase fixes: Bernabé’s finding is a near-absence of empirical evidence (one surveyed study tested the claim, no significant effect), not “no measurable benefit”; and the unverifiable “12,288 OWL approximations” figure was softened to the qualitative claim (neither is in the gUFO abstract).
3. Stale corpus reference noted (text stands). “35 of 47 opda:ufoCategory tags in opda-vocabularies.ttl” describes the session-040 deliberation-time state. ADR-0044 Phase 5c then promoted the facet onto classes, and ODR-0031/ADR-0045 relocated all tags to opda-annotations.ttl (≈40), retyped owl:AnnotationProperty, and split the register-deference axis off the predicate — implementing this record’s Rule 2. opda:ufoCategory now carries only the genuine UFO class categories; “Quale-in-Region” is no longer one of its values. The normative text stands; this is a provenance pointer.
4. Held-dissent provenance — trigger (i) fired and was cured. Cagle’s re-open trigger (i) (“the ufoCategory layer ever made reasoned-over”) fired via ADR-0044 Phase 5c (tags landed in reasoned-union graphs) and was remediated by ODR-0031/ADR-0045 (relocation + owl:AnnotationProperty + the sixth three-graph CI gate). The held dissent remains live on the optional-vs-load-bearing question.
5. Overclaim clarified (aptness, not uniqueness). “Only UFO+UFO-L unifies the reification primitive with Hohfeldian correlativity” is read as aptness among the practical options (the reification substrate is already UFO’s Relator), not metaphysical uniqueness — LegalRuleML and plain Hohfeld-in-SHACL also formalise correlativity. The page wording was softened.

Engineering follow-ups routed (regression-hardening; → ADR-0045, not blocking): (a) widen the sixth three-graph CI gate to the full ODR-0029 reasoned union — it scans 8 of 10 graphs, omitting opda-vocabularies.ttl + opda-contexts.ttl (one of which the original breach reached); (b) add a TBox-only test that the opda:ufoCategory sh:in meta-shape never acquires a domain-class target. Process note: external citations need web-verification, not persona judgment — session-040’s pre-flight “citations grounded ✓” covered corpus facts but not the external bibliography; split the two gates.

2026-06-15 — RATIFIED proposed → accepted (operator)

The operator ratifies adoption. Basis: the session-044 validation pass above found the decision sound / complete / consistent and unchanged, the corrections were page/provenance-only (no normative change), and the two regression follow-ups it surfaced are themselves shipped + green (ADR-0045 §Follow-ups; the three-graph gate is now seven checks). Ratified jointly with its enforcing/realising records ODR-0031 + ADR-0045. The council shapes the proposal; this records its adoption.

More Information

• Council transcript: session-040 — The foundational-ontology choice (full per-question dialectic, the OntoClean≠UFO wedge, the Smith↔Guizzardi Q3 settlement, tally + DA scorecard).
• Enforced + realised by: ODR-0031 + ADR-0045 (council session-041) — Rule 1 (annotation-graph-only) and Rule 7(b) (DOLCE disclosure) were found breached in the committed corpus by ADR-0044 Phase 5c; ODR-0031/ADR-0045 restore the quarantine, retype opda:ufoCategory owl:AnnotationProperty, and add a sixth three-graph CI gate so the breach cannot silently re-ship.
• Reference page: /ontology/foundational-ontology (src/pages/ontology/foundational-ontology.astro) — the sourced critique, the alternatives comparison, and the verdict this ODR records.
• Depends on / scoped by: ODR-0029 (the inference/validation boundary + gUFO quarantine), ODR-0011 §8a (the seven-category framework), ODR-0027 (classification-over-inheritance), ODR-0010 §Q7a + ODR-0004 §3a (annotation-graph home / three-graph separation); ADR-0034 (the gated gUFO typing pass).
• External citations: Almeida, Guizzardi, Sales & Fonseca, gUFO: A Gentle Foundational Ontology (2026) — the OWL-2-DL undecidability warrant; Bernabé et al., Journal of Biomedical Semantics (2023) — no-measurable-benefit; Merrill, Applied Ontology (2010) — BFO’s benefits do not require its realism; Griffo, Almeida & Guizzardi — UFO-L legal relations; Masolo et al., WonderWeb D18 (2003) — the DOLCE quality structure.