The Knowledge Graph Explained
Session 1.3 · ~5 min read
Search for "Joko Widodo" on Google. On the right side, you see a Knowledge Panel: his photo, title, birth date, spouse, political party, education, and related people. Google did not crawl the web in real time to assemble this panel. It pulled pre-stored facts from its Knowledge Graph, a massive internal database of entities and their relationships.
The Knowledge Graph is not a search index. The search index stores web pages. The Knowledge Graph stores things: people, companies, places, products, concepts. Each thing has properties (attributes) and connections to other things (relationships). When Google answers a factual question or displays a Knowledge Panel, it queries this graph, not the web.
Scale and Structure
Google's Knowledge Graph contains over 500 billion facts about over 5 billion entities. Each entity has a unique machine identifier and a set of typed relationships to other entities. The structure is a directed graph: nodes are entities, edges are relationships.
(Person)"] -->|"president of"| RI["Republic of Indonesia
(Country)"] JW -->|"born in"| SK["Surakarta
(City)"] JW -->|"political party"| PDI["PDI-P
(Political Party)"] JW -->|"spouse"| IR["Iriana Joko Widodo
(Person)"] RI -->|"capital"| JKT["Jakarta
(City)"] SK -->|"located in"| JT["Central Java
(Province)"] style JW fill:#2a2a28,stroke:#c8a882,color:#ede9e3 style RI fill:#2a2a28,stroke:#6b8f71,color:#ede9e3 style SK fill:#2a2a28,stroke:#6b8f71,color:#ede9e3 style PDI fill:#2a2a28,stroke:#8a8478,color:#ede9e3 style IR fill:#2a2a28,stroke:#8a8478,color:#ede9e3 style JKT fill:#2a2a28,stroke:#6b8f71,color:#ede9e3 style JT fill:#2a2a28,stroke:#8a8478,color:#ede9e3
Every node in this graph is an entity. Every arrow is a relationship. The Knowledge Graph is this structure, replicated billions of times across every domain of human knowledge.
The Knowledge Graph is not a list of facts. It is a web of connected entities. The more connections your entity has, the more confidently Google recognizes it.
Where the Data Comes From
The Knowledge Graph draws from multiple sources. Understanding these sources reveals how to get your entity into the graph.
| Source | Type | Contribution | Your Action |
|---|---|---|---|
| Wikidata | Structured database | Entity properties, relationships, identifiers | Create/maintain entry |
| Wikipedia | Encyclopedia | Entity descriptions, notability confirmation | Qualify through reliable sources |
| Google Business Profile | Google property | Local business entities, location, categories | Claim and optimize |
| Schema.org markup | Website structured data | Entity declarations, properties, relationships | Implement on your site |
| Authoritative websites | Third-party sources | Corroboration, mentions, citations | Build presence on directories |
| Government databases | Official records | Legal entity confirmation | Ensure registration is current |
No single source is sufficient. Google cross-references multiple sources to build confidence in an entity's existence and attributes. This is why entity infrastructure requires presence across multiple platforms, not just a well-built website.
How the Graph Powers Search Features
The Knowledge Graph feeds several visible search features. Each feature represents a different way Google uses entity data:
- Knowledge Panels: The information box on the right side of search results for recognized entities. Pulled directly from the Knowledge Graph.
- AI Overviews: Google's AI-generated answers draw entity facts from the Knowledge Graph to compose responses.
- Local Pack: The three-pack of local businesses shown for location-based queries. Powered by GBP data in the Knowledge Graph.
- Rich Results: Star ratings, FAQ dropdowns, breadcrumbs, and other enhancements triggered by schema markup that connects to entity data.
- People Also Ask: Questions related to entities in the query, answered using Knowledge Graph relationships.
- Disambiguation: "Did you mean Apple Inc. or apple fruit?" powered by entity type differentiation in the graph.
If your business is not in the Knowledge Graph, it cannot appear in any of these features. You are limited to organic blue links, competing purely on content signals.
The Graph Grows Through Reconciliation
Google does not simply add entities to the Knowledge Graph. It reconciles them. When Google encounters information about a company on your website, on your GBP, and on a directory, it must determine: are these the same entity? If the name, address, and other signals match, Google merges them into a single entity node. If they do not match, Google may create separate nodes or simply discard the ambiguous data.
This reconciliation process is why consistency across all platforms matters. Inconsistent information does not just look sloppy. It prevents Google from building a unified entity for your business.
The chart above is an estimated distribution. Google does not publish exact weights. But the hierarchy is well-established through practitioner research and patent analysis: Wikidata and Wikipedia carry the most authority, followed by Google's own properties, then structured data, then third-party corroboration.
Further Reading
- Google Knowledge Graph Reconciliation - Bill Slawski's analysis of how Google reconciles entities across sources
- Understanding Google Knowledge Graphs - Overview of Knowledge Graph structure and function
- What Is the Google Knowledge Graph? - Practical explanation of how the Knowledge Graph impacts search visibility
Assignment
Search for any famous person (e.g., "Joko Widodo"). Look at the Knowledge Panel on the right. List every piece of information shown: name, title, born, spouse, education, and so on. Each of these is a property stored in the Knowledge Graph. Your company needs the same kind of structured presence. Count how many properties are displayed and compare that to how many properties Google currently stores about your business.