Knowledge Graph

A knowledge graph stores facts as entities connected by typed relationships.

People, places, products, and concepts are nodes. Edges are relations such as "works for," "located in," or "produces." Attributes, timestamps, and confidence scores sit on the graph so queries can use context, not isolated facts. Google's Knowledge Graph is the public example: search a person and the sidebar of career, birthdate, and related people comes from the graph.

Businesses use graphs so search and recommenders know a product belongs to a category, a customer bought similar items, and those items share a brand. Enterprise graphs link customer records, support tickets, and contracts so staff can trace a problem across systems without stitching tables by hand.

Healthcare graphs connect patient records, papers, and treatment guidelines to speed diagnosis and drug discovery.

A node is an entity. An edge is a typed fact, optionally with time and a confidence score. Queries can hop: person works-for company located-in city. Google’s Knowledge Graph fills the search sidebar with career, birthdate, and related people from that structure. Commerce graphs encode category, brand, and purchase history for search and recommendations.

Enterprises link customers, tickets, and contracts so a root cause is a graph walk, not a join marathon. Clinical graphs link records, literature, and guidelines for diagnosis and drug discovery. Google's 2012 Knowledge Graph post: search entities and relations, not only strings. Wikidata is the public cousin.