Knowledge graph
Also known as: Google Knowledge Graph, Knowledge panel
A knowledge graph is a data structure that stores entities such as people, organizations, places and concepts, along with the relationships between them, as a connected network; Google's Knowledge Graph, introduced in 2012, is the best-known example.
In a knowledge graph, each node is an entity and the lines between nodes are relationships: “Company X is based in City Y”, “Person Z founded Company X”. The knowledge panels Google shows beside search results are drawn from its graph. Wikidata is an open knowledge graph anyone can access.
How do you get into it?
Google doesn’t take direct submissions to its Knowledge Graph. It compiles information from web sources, licensed data and structured data. Once a knowledge panel exists for a business, an authorized person can claim it and suggest corrections. Things that make inclusion easier:
- Marking up organization and people details with structured data
- Using exactly the same name, address and phone number everywhere
- Being mentioned in independent, trusted sources
How it relates to AI
Large language models don’t store knowledge as a graph; they hold patterns learned during training. Even so, companies like Google use knowledge graphs alongside their search and AI products, though the details aren’t public. A brand that’s clearly defined as an entity is less likely to be mismatched on either side.
Example
If searching for a manufacturer’s name brings up a panel on the right with its founding year, headquarters and website, Google recognizes that company as an entity in its Knowledge Graph.