A knowledge graph represents facts as semantic triples: (Subject, Predicate, Object) — e.g., (Einstein, bornIn, Ulm). Nodes are entities (people, places, concepts); edges are typed relations. RDF (Resource Description Framework) is the standard format; SPARQL is the query language. Large-scale examples include Wikidata (100M+ triples), Google Knowledge Graph, and Freebase. The highlighted subgraph shows all entities reachable within the selected number of hops from the query node (gold) — the basis of multi-hop question answering in modern AI systems.