Relational Verification for AI Indexing: Schema.org, OAI-PMH, and JSON-LD Extensions for Metadata Packet Infrastructure

This proposal argues that AI-indexing standards require relational verification fields in addition to cryptographic provenance. Decentralized identifiers (DIDs) can verify control of a private key associated with an identifier; semantic-integrity proofs can verify formally declared constraints; validator networks can detect some evidence conflicts and gaming behavior. None of these mechanisms, by itself, verifies whether an entity's self-description is socially, historically, or authorially warranted. This gap becomes visible in the ontological-claimant problem: cases where an entity benefits from inaccurate indexing, where self-inscription depends on retrieval-layer compression damage, where the entity is uncooperative with accurate verification. We do not propose a fourth epistemic-adjudication layer. We propose standardizing the relational grammar already specified in the Metadata Packet for AI Indexing (MPAI; DOI: 10.5281/zenodo.19578086): entity definitions, disambiguation matrices, negative tags, semantic integrity markers, typed DOI reference lists, evidence membranes, and depth-proof metrics. Three concrete additions to existing standards infrastructure: Schema.org Extension Vocabulary — namespace mp: with full term definitions, cardinalities, and per-claim verification structure (replacing the single verification-level ladder with mp:claimStatus + mp:verificationSignals + mp:ontologicalStatus) OAI-PMH Metadata Profile — oai_mpai with explicit metadataPrefix, namespace, and schema specifications, designed for trivial mapping to ResourceSync and Signposting JSON-LD Vocabulary — stable URI at https://metadatapacket.org/vocabulary The additions are non-displacing and individually adoptable. Adoption sequence: vocabulary publication 0–3 months, Schema.org community proposal 3–12 months, OAI-PMH profile registration 12–24 months, NIST/ISO/W3C alignment 24–60 months. Includes Security Considerations (§IX), IANA Considerations (§X), and a worked example (Appendix B) demonstrating how an ontological-claimant case structures in MPAI to expose its own self-asserted nature without requiring protocol-level adjudication. Central claim: The MPAI grammar verifies relationships, not assertions. The protocol does not decide whether a claimant is sincere, artistic, mistaken, or fraudulent — it exposes which claims are self-asserted, which are corroborated, which are disputed, which are excluded, and which relationships support or fail to support them. Adjudication remains with the citation environment; the protocol makes the structure legible. Document classification: EA-WL-01 v1.1 · Phase X Standards-Track · Working Draft draft-sharks-mpai-relational-01Hex: 06.SEI.STANDARDS.RELATIONAL.01Companion deposit: Metadata Packet for AI Indexing (DOI: 10.5281/zenodo.19578086).∮ = 1