G.13:12 - SoTA‑Echoing (post‑2015, for orientation; non‑normative)

• Scholarly claim graphs & open indexes. Open research KGs and open scholarly indexes encourage claim‑level representations and concept taxonomies as interop substrates (post‑2015 ecosystem: KG‑style contribution graphs; open indexing initiatives). Treat these as sources registered via ExternalIndexCard, not as governing patterns.

• Neural representations for scientific text. Transformer‑based scientific encoders (e.g., SciBERT‑class; citation‑aware paper representations such as SPECTER‑class; later retrieval‑oriented scientific embedding families) are useful as alignment heuristics. In FPF terms, they belong behind ScaleEmbeddingSpec + pinned editions/policies (see G.13:Ext.EmbeddingBasedAlignment).

• Schema matching & entity resolution (deep‑learning era). Modern matcher families (deep entity matching, contrastive representation alignment, GNN‑assisted graph alignment) help populate interop cards, but must not become “implicit semantics”; record their use as policy‑bound wiring in extensions.

• Systematic review process modernisation. PRISMA‑2020‑class review records (post‑2015 practice) are valuable as evidence anchors and coverage telemetry; treat them as evidenced inputs (EvidenceGraph anchors + pinned editions/windows), not as legality gates.

• QD / Illumination and OEE declared set results. Post‑2015 QD (MAP‑Elites successors, CMA‑ME line, differentiable QD toolkits) and OEE (POET‑class and related environment/method co‑evolution lines) often rely on external taxonomies and environment corpora. Interop should expose those as pinned external editions and keep coverage/regret as telemetry inputs—never as implicit dominance.