Skip to main content

Knowledge Graph

A machine-readable map of facts about the world, connecting entities like companies, people, and products to each other and their attributes in a structured, queryable way.

The Simple Version

A machine-readable map of facts about the world, connecting entities like companies, people, and products to each other and their attributes in a structured, queryable way.

Detailed Explanation

Knowledge graphs store factual information as subject-predicate-object triples (e.g. 'Company X, headquartered in. City Y') in a graph data model. They power entity disambiguation (determining which 'Apple' is meant, the company or the fruit), fact verification, and structured question answering. Commercial examples include Google Knowledge Graph, Wikidata, and DBpedia. In AI applications, knowledge graphs reduce hallucination by grounding entity-related claims in structured, maintained fact stores. Enterprise knowledge graphs for product catalogues, employee directories, and regulatory taxonomies can significantly improve AI retrieval accuracy for domain-specific applications.

Key Characteristics

  • Stores facts as typed entity-relationship-entity triples
  • Enables entity disambiguation, relationship traversal, and structured querying
  • Used for grounding, fact verification, and structured reasoning
  • Enterprise knowledge graphs improve AI retrieval accuracy for domain-specific applications

Why It Matters

Organisations building authoritative knowledge graphs for their domain (products, regulations, company entities) create retrievable structured knowledge that AI systems can use more reliably than unstructured text, improving retrieval quality and reducing hallucination.

Real-World Analogy

Like a detailed, interconnected encyclopaedia where every article explicitly cross-references related articles with labelled relationship types, enabling a reader (or AI) to navigate and reason across the knowledge base systematically.

Common Misconceptions

  • Knowledge graphs and relational databases are equivalent, knowledge graphs support flexible, semantically typed relationships without a predefined schema, enabling richer reasoning than relational models.
  • Building a knowledge graph requires starting from scratch, organisations can bootstrap knowledge graphs from existing structured data sources (databases, product catalogues, taxonomies).

Related Terms

Related Articles

Sources & Further Reading