Root-LD
Federated Semantic Linked Data Specification
Every entity in a Root-LD federation, regardless of domain, regardless of type, regardless of content, carries the same root wrapper. The root never changes. The content beneath it does. The edges between entities emerge from the data over time through deterministic, lexical, and semantic passes.
The result is a knowledge graph that grows more intelligent as it grows larger — because every new entity added to any domain in the federation immediately becomes available for edge-building against every existing entity across all domains.
Root-LD is not a content management system. It is not a database schema. It is not an SEO framework. It is a semantic architecture for federated knowledge that happens to produce all three as byproducts.
Every Root-LD entity consists of exactly three layers stacked in fixed order. The layers never reorder. The layers never merge. Each layer has a specific function and a specific set of fields. Some fields are required. Some are conditionally required. Some are optional. None are invented at the entity level — all fields are defined in the specification and populated from data.
Layer One — Root-LD Anchor
FIXED · UNIVERSALThe anchor layer. Fixed across all entities across all domains across all time. The only field that varies between domains is the domain signature. Every other field in this layer is either universal or derived deterministically from the entity's properties at ingestion time.
Purpose: To make every entity in the federation uniquely identifiable, verifiable, traceable to its primary source, and locatable within the federation hierarchy — regardless of what domain it lives in or what type of content it contains.
Anchor Fields
LAYER ONE// Root-LD Anchor — Layer One "specVersion": "1.0", "federationId": "OM-STATUTE-7x9k2m", "domainSignature": "oakmorel.com", "entityClass": "STATUTE", "entitySubclass": "FEDERAL-USC", "primarySource": "https://uscode.house.gov/view.xhtml?req=15+U.S.C.+1", "sourceVerified": true, "dateIngested": "2026-02-28T00:00:00Z", "dateLastVerified": "2026-02-28T12:00:00Z", "dateModified": "1890-07-02T00:00:00Z", "contentHash": "sha256:a1b2c3d4...", "humanVerified": false, "generationMethod": "LLM-ASSISTED", "disclosureStatement":"This summary was generated by an automated machine intelligence system. It is not a legal interpretation. The original source text is preserved verbatim below."
Layer Two — Body-LD
DOMAIN-SPECIFIC · CONTENTThe content layer. Domain-specific and entity-class-specific. This is where the actual data lives. The body schema is defined per entity class at the domain level but must conform to federation-level field naming conventions so that cross-domain edge-building can operate on shared field names without custom parsers for every domain.
Universal Body Fields
ALL ENTITIES · ALL CLASSESPresent in every entity regardless of class or domain.
Class-Specific Body Fields
PER ENTITY CLASSDefined per entity class at the domain level. Selected classes shown below. All class-specific fields must conform to federation-level naming conventions.
"citationFormal": "15 U.S.C. § 1", "citationInformal": "Sherman Antitrust Act Section 1", "amendmentHistory": [ ... ], "relatedStatutes": [ "OM-STATUTE-...", "OM-STATUTE-..." ], "regulatoryAuthority": "Federal Trade Commission", "penaltyProvisions": true, "procurementRelevance": false
"solicitationNumber": "FA8501-26-R-0001", "issuingAgency": "OM-ENTITY-af7b2c", "naicsCode": "541330", "setAsideType": "SDVOSB", "estimatedValue": 4200000, "responseDeadline": "2026-04-15T17:00:00Z", "awardedTo": null, "awardAmount": null, "contractType": "FIXED-PRICE"
"indicatorName": "Consumer Price Index — All Urban Consumers", "indicatorCode": "CUUR0000SA0", "issuingAgency": "U.S. Bureau of Labor Statistics", "reportingPeriod": "2026-01", "value": 315.8, "unit": "Index (1982-84=100)", "priorPeriodValue": 314.2, "methodology": "https://www.bls.gov/cpi/methodology.htm"
Layer Three — Recursive-LD
EDGE LAYER · GROWS OVER TIMEThe edge layer. This is where the graph becomes intelligent. Recursive-LD is not populated at ingestion. It grows over time through sequential passes run against the full corpus.
It is called Recursive-LD because each pass uses the output of prior passes as input — the graph reasons about itself using its own prior reasoning as context.
Edge Object Structure
LAYER THREEFederation Edge Taxonomy
DEFINED RELATIONSHIP TYPESThe defined set of relationship types across the federation. Edge types not in this taxonomy cannot be created without a specification amendment.
The Federation Passes
EDGE BUILDING · SEQUENTIALFour sequential passes run against the full corpus to populate the Recursive-LD layer. Each pass uses the output of prior passes as input.
Registered Domains
CURRENT FEDERATIONEach domain in the federation has a registered profile defining its primary entity classes, domain lexicon contribution, schema extensions, and relationship to other domains.
Edge Density — The Primary Metric
COMPOUND REASONINGThe power of the federation is measured not by the number of entities but by the density of cross-domain edges. A graph with 10,000 entities and 1,000 cross-domain edges is less powerful than a graph with 1,000 entities and 10,000 cross-domain edges.
Edge density is the metric that matters. Every cron run, every new entity ingested, every LLM pass increases edge density. The graph compounds.
Immutability Rules
NON-NEGOTIABLEThese four fields ensure that any external system that has recorded a federationId can always resolve it and trust what it finds.
Deprecation Protocol
ENTITIES NEVER DELETEDEntities are never deleted. They are deprecated. A deprecated entity receives status: SUPERSEDED or status: EXPIRED, a supersededBy field pointing to the new entity's federationId, and is retained in the graph permanently.
This ensures that any external citation of a deprecated entity resolves to a page that explains what superseded it and where the current version lives.
Version Control
BACKWARD COMPATIBLEThe specification version is recorded in every entity's anchor layer. When the specification changes, a new version is issued. Existing entities retain their original spec version. The federation maintains backward compatibility across all versions.
The following terms are canonical within the Root-LD specification and across all federation domains. Use these names — not synonyms, not abbreviations.