BKG-1.0: Brand Knowledge Graph
Model entities, aliases, relationships, claims and evidence to build a governed semantic representation of the brand.
How to interpret this document
This content describes technical and methodological behavior that is implemented or explicitly planned in the product. When a control depends on configuration, a provider, a secret, a contract or legal approval, that dependency must remain visible.
Purpose
BKG-1.0 creates a canonical semantic layer for organizations, brands, products, websites, markets, competitors, attributes, channels and related topics.
Entities and aliases
Each entity has a canonical key, type, name, description, market, locale, ownership and confidence. Aliases map abbreviations, former names, domains, handles and identifiers without duplicating entities.
Relationships
The graph uses explicit relationships such as owns, represented_by, operates_in, competes_with, has_attribute and related_to. Relationships can be candidate, verified, rejected or archived.
Claims and evidence
Claims capture verifiable statements about identity, ownership, offering, market, attributes or positioning. Evidence preserves source, URL, excerpt, weight, hash and verification state.
Governed ingestion
Crawler Intelligence, Source Intelligence and Prompt Demand can create KnowledgeGraphCandidate records. Those signals do not enter the canonical graph automatically; review is required before promotion.
Health scores
BKG-1.0 calculates Identity, Consistency, Evidence Coverage, Relationship Verification and Coverage. These indicators measure internal completeness and governance, not whether external AI models understand the brand identically.
Contamination protection
Contested claims, ambiguous aliases and unsupported relationships remain outside the verified layer until reviewed. Identity and ownership require explicit evidence in the baseline.
Next layer
Narrative & Perception Intelligence uses the Knowledge Graph as canonical vocabulary to compare what the brand declares with what models and external sources say.