Knowledge graphs are structured representations of information where entities (nodes) are connected by relationships (edges), often in the form of triples: (subject, predicate, object). They model real-world knowledge in a way that’s both human-readable and machine-actionable.

Used in search, recommendation systems, question answering, and data integration, knowledge graphs enable reasoning over data and context-aware inference. For example, a query like “films directed by Christopher Nolan” can be answered by traversing relationships in a graph.

They can be built from curated data, extracted via NLP from unstructured text, or derived from existing databases. Popular examples include Google’s Knowledge Graph, Wikidata, and DBpedia.

Key challenges include entity resolution, relationship extraction, and ensuring graph consistency. Machine learning methods like graph embeddings and graph neural networks (GNNs) are often used to operate on or learn from knowledge graphs.