A structured network of entities and their relationships stored as nodes and edges; the explicit kind powers Google's knowledge panels, and an LLM carries an implicit statistical version in its weights.
Knowledge Graph is a structured network of entities and the relationships between them: people, places, organisations, products, and concepts stored as nodes, with typed edges recording how they connect. It represents facts in a form a machine can traverse and reason over, rather than as free text.
The best-known example is the Google Knowledge Graph, which powers the knowledge panels in search and supplies the canonical facts a search engine shows about an entity, such as its founder, headquarters, or parent company. Wikidata, Wikipedia infoboxes, and schema.org markup feed graphs like it, which is why consistent structured data and third-party corroboration help a brand resolve to a single entity rather than an ambiguous string.
An LLM does not query an external knowledge graph the way search does. It carries an implicit, statistical version in its weights: the web of associations it learned during training. We map that hidden structure through model probing, recording the entities, competitors, and attributes a model links to a brand as its brand association network. Strengthening a brand's place in both the explicit graph and the implicit one is a core part of AI visibility work.
