Schema Implementation for Knowledge Entities

Schema markup is the technical layer that transforms content from a document into a knowledge graph node. Without it, even the most precisely defined concept remains a text file from the knowledge graph’s perspective. With it, the concept becomes an entity — something the graph can associate with a creator, link to adjacent concepts, and present as an authoritative answer.

The minimum viable schema set for a knowledge entity

For a concept defined in the Ignorance Graph methodology to achieve full entity status, it requires three schema types working in combination:

1. WebSite schema (on the root domain)

Establishes the site as a named entity with a stable identifier. Required for all sameAs relationships to function correctly. Must include: @id (canonical URL), name, url, and potentialAction (SearchAction).

2. Person or Organization schema (on the /about/ page)

Establishes the creator as a named entity associated with the site and all concepts defined on it. Must include: @id, name, url, sameAs (LinkedIn, publisher profiles, Wikidata Q-ID when available). This is the entity that the knowledge graph will associate with all DefinedTerms created on the site.

3. DefinedTerm schema (on each definition page)

The core entity markup for each concept. Must include: @type: DefinedTerm, @id (canonical URL of the definition page), name (the exact term), description (the precise definition, 1–3 sentences), inDefinedTermSet (reference to /glossary/), and termCode (a stable unique identifier for the concept).

sameAs and external confirmation

Schema markup alone creates an unconfirmed entity. External confirmation — a Wikidata Q-ID, a reference in an authoritative external source — transforms an unconfirmed entity into a confirmed one. The sameAs property in DefinedTerm schema is the technical slot for this confirmation. Leave it as a placeholder until external references exist; do not fill it with incorrect URLs.

Deployment timing

Schema should be deployed fully — with all required properties present — or not at all. Partial schema with placeholder values or incorrect URLs is worse than no schema: it creates an entity assertion that the knowledge graph cannot confirm, which can delay entity recognition rather than accelerate it.

The correct sequence: establish the logo URL, canonical URLs, and at least one external sameAs target — then deploy schema across all pages in a single pass. See the Ignorance Graph Analysis deliverable for the schema templates included in the architecture document.