ν.β E1 — Tier 2 retrieval coverage backfill — CLOSED 2026-05-04

Outcome

outcome-VALIDATED-WITH-NOTE

Kill condition NOT-MET — gaps are SYSTEMATIC, not random outliers.

Key findings

  1. 102 library/indexed/ books = 0 pgvector chunks — root cause of G1/G3/G4 gaps. The Tier 2 DB (7963 total chunks) contains only T-files, v6.6 schema files, ε.ι T-files, and cumulative-state docs. No book content is indexed.

  2. Tarjan 2024 DOI 404 — HALLUCINATION confirmed — DOI 10.1016/j.websem.2024.100806 returns HTTP 404. First flagged in Z1 v1.0 honesty_caveats. E1 independently confirms. Spike E2 must NOT cite Tarjan 2024; must survey actual incremental OWL EL literature.

  3. G1 ODP role-modelling: systematic gap. Z3’s 0/6 was accurate at run-time. Current DB returns only tangential T-file noise (sim 0.636-0.771). keet-ontology-engineering-1e, hitzler-foundations-semweb, hogan-knowledge-graphs are on disk but not indexed.

  4. G3 IRI-registry: partial — T7-AKOMA-NTOSO covers ELI/ECLI (sim=0.773). No w3id/identifiers.org/PURL coverage. eu-eli-technical-guide + eu-ecli-specification on disk, not indexed.

  5. G4 cross-module: partially covered — ε.ι S2 T-file + T13 + T44 (sim 0.779-0.832). Best-covered gap of the 4.

Backfill action plan (summary)

Priority 1 — ingest already-owned books (£0 / ~½d):

  • keet-ontology-engineering-1e (G1)
  • hitzler-foundations-semweb (G1)
  • hogan-knowledge-graphs (G1/G3/G4)
  • allemang-semantic-web-3e (G3/G4)
  • labra-gayo-validating-rdf-data-2e (G4)
  • eu-eli-technical-guide (G3)
  • eu-ecli-specification (G3)
  • shimizu-hitzler-llm-oe-2025 (G1 adjacent)

Priority 2 — free-PDF acquisitions (~1-2h):

  • Kazakov et al. ELK JAR 2014 DOI 10.1007/s10817-013-9296-3 (G2 replacement)
  • Grau et al. JAIR 2008 modular ontology (G4)
  • W3C Cool URIs note (G3)

Process fix: add pre-suite ingestion check to spike-suite pre-flight checklist.

T-file

/home/richardd/off-github/library/projects/inherit/T-spike-nu-beta-E1-tier-2-retrieval-coverage-backfill-2026-05-04.md