Knowledge Graph
A structured representation of real-world entities and relationships, stored as a network of nodes (entities) connected by edges (relationships).
Structure
Knowledge graphs store facts as triples: (subject, predicate, object). Example: (Albert Einstein, born_in, Ulm). Entities are nodes, relationships are edges. This creates a queryable, interconnected web of knowledge.
Examples
Google Knowledge Graph (powers search info boxes), Wikidata (open knowledge base), DBpedia (extracted from Wikipedia), and enterprise knowledge graphs used in healthcare, finance, and e-commerce.
Knowledge Graphs + LLMs
GraphRAG combines knowledge graphs with retrieval-augmented generation. The structured relationships in knowledge graphs help LLMs reason about entities, resolve ambiguity, and produce more accurate, grounded answers.