project_zeta_3_Z2_vocabulary_management_2026_05_03

ν.α/ζ.3 Spike Z2 — Vocabulary management 2025-2026 SOTA — CLOSED 2026-05-03

Outcome: outcome-VALIDATED

Target Q-NU: Q-NU-002 (vocabulary management strategy) Target ζ-Q: ζ-Q7 (faceted classification re-ask)

Why: Killed the question of whether flat vs faceted SKOS is correct for Phase-1 INHERIT v2 vocabulary management. Pre-derisked substrate for ζ-Q7 re-ask under refined-prompt v3.8.

How to apply: When ζ-Q7 is formally asked, Q-NU-002-vocabulary-management-strategy.md has §0-§9 already populated with empirical evidence from 5 production deployments (AGROVOC, EuroVoc, GettyAAT, LCSH, Mondo) + 3 programmatic scenarios (rdflib + skosify validated, exit 0). No re-derivation needed.

Key findings

1. Phase-1 posture: Scenario A — Flat SKOS (single ConceptScheme, ~1K concepts, jurisdiction-tagged via skos:scopeNote). Q-005 κ.δ compatible.

2. SkoHub SHACL-Actions (April 2024) is the correct Phase-1 SKOS CI gate. Direct GitHub Actions plug-in. Zero custom CI work.

3. Critical governance discipline: explicit skos:topConceptOf on 5-10 root concepts + skos:broader chains on all non-root. Empirically validated — skosify 2.3.0 auto-promotes all 1,000 flat concepts to topConceptOf when hierarchy is absent (1,000 INFO messages, exit 0).

4. SKOS-XL: NOT needed for Phase-1. Community view: “most vocabularies don’t need it” (T5/TopQuadrant). Multilingual Phase-2+ only.

5. EuroVoc finding: “faceted” grouping in EuroVoc uses skos:broader hierarchy within a single flat ConceptScheme — NOT multiple ConceptSchemes per theme. Scenario B multi-scheme approach is therefore an unnecessary complexity vs established practice.

6. Phase-1.5+ migration path: Scenario C (hybrid flat + 3 per-faith-pillar facets). Triggers at faith-pillar wave-1 (per 22-spike S6 root-cause).

7. Kill condition NOT-MET: 0/5 deployments reveal fundamental tooling unsupported-ness. All tooling (SkoHub, VocBench, skosify, Mondo SSSOM) supports all 3 scenarios.

Artefacts

• T-file: ~/off-github/library/projects/inherit/T-spike-zeta-3-Z2-vocabulary-management-2026-05-03.md (v1.0)
• Q-NU-002: ~/testatetech/docs-strategy/docs/superpowers/specs/2026-04-29-multi-phase-audit/current-questions/Q-NU-002-vocabulary-management-strategy.md (v0.2; state 3)
• Programmatic scenarios: /tmp/spike-zeta-3-Z2/scenario_A.ttl, B.ttl, C.ttl (1K concepts each; skosify exit 0)

Scenario evaluation matrix (summary)

Scenario	Label	Total effort est	Governance overhead	Phase fit
A	Flat SKOS (κ.δ baseline)	750h / 18.8 pw	LOW	Phase-1 RECOMMENDED
B	Faceted per-jurisdiction	800h / 20.0 pw	MEDIUM-HIGH	Year-2+
C	Hybrid + faith-pillar facets	825h / 20.6 pw	LOW	Phase-1.5+

Richard-tasks candidates (for REFERENCE session)

• Add SkoHub SHACL-Actions as A-21 CI gate candidate (#23 or next available)
• Author skos:topConceptOf + skos:broader governance rule in code-inherit-v2 BUILD-PLAN
• Register INHERIT v2 vocabulary schemes in BARTOC + FAIRsharing at Phase-1.5+ launch

T-files consulted

• T5-SKOS-XKOS-CMNS-notes.md (primary anchor; Tier 2 sim 0.72-0.76)
• T18-BIOLINK-MODEL-notes.md (governance precedent)
• T82-architecture-evaluation-methodology-2026-04-30.md (κ.θ utility-tree)