Glossary /Structured data
Knowledge Graph
A search engine’s database of entities and the relationships between them, used to answer queries about things rather than strings.
A knowledge graph stores entities — people, companies, places, concepts — and how they relate, rather than storing documents that mention them. It is how an engine knows that a company, its founder and its product are three distinct things with connections between them, and how it answers a question without ranking a page.
Entry is by corroboration. An engine adds an entity when independent sources agree on it: consistent naming across a company’s own site, structured data that states the relationships explicitly, and third-party references that confirm the same facts. One source asserting something about itself is not enough.
This has become considerably more consequential with assistants in the picture, since a generated answer about a company draws on the entity the engine has assembled rather than on whichever page currently ranks first.