A structured network of real-world entities and their relationships used to organise and query connected information.
Also known as: KG, semantic network
In plain English
A knowledge graph is like a map that shows things and how they connect, people, products, places and facts. It helps search and assistants find related information and explain why results are linked, working quietly behind the scenes.
In practice
In practice, knowledge graphs provide canonical entities and relationships that power search, recommendations and chatbots, improving accuracy, disambiguation and explainability. They matter because they let teams integrate data across systems, enforce a shared schema and enable traversable queries for business logic and audits.
Under the hood
A knowledge graph models facts as nodes (entities) and edges (relations) with typed properties and an ontology or schema; common forms include RDF triples and property graphs. Implementations use graph databases and unique identifiers (URIs), queried via SPARQL or Cypher, with pipelines for entity extraction, linking, and optional integration with embeddings or reasoning engines for semantic retrieval and inference.
Example
"Can we add the product taxonomy and customer IDs to the knowledge graph so the chatbot returns accurate, explainable answers?"
Related terms
Vector Database · Embedding · Semantic Search · Retrieval-Augmented Generation