A structured map of things and how they relate: this company employs this person, who wrote this paper, which cites that one. Unlike a model’s diffuse learned knowledge, a knowledge graph is explicit, checkable and editable, which makes it valuable precisely where language models are weak. Combining the two is an active and useful area.
Technical view
Entities and relations stored as triples and queried with languages such as SPARQL or Cypher over graph databases. Construction is the hard part: entity resolution, schema design and keeping the graph current. In retrieval-augmented systems, graph-based approaches improve multi-hop questions where the answer requires joining facts that appear in different documents, which flat vector search handles poorly. The trade-off is construction and maintenance cost against precision and explainability, and hybrid designs that retrieve both text and graph context are common.
Related
Part of AI Terminology Index, section: Adjacent Fields.