Your First Knowledge Graph

What a Knowledge Graph Actually Is

A knowledge graph (KG) stores information as a network of things and the connections between them, instead of as rows in tables. If you have ever sketched boxes-and-arrows on a whiteboard to explain a system, you have already drawn a knowledge graph.

The Three Primitives

Every knowledge graph is built from exactly three kinds of things:

1. Nodes (also called vertices or entities) — the things you care about: a person, a company, a product, a city, an order.
2. Edges (also called relationships) — a typed, directed connection between two nodes: Alice works_at ACME. The direction matters: works_at points from the person to the company, not the other way around.
3. Properties — key/value attributes attached to a node or an edge: a person node has name: "Alice" and born: 1995; the works_at edge has since: 2021.

Why Engineers Reach for Graphs

The real world is relationship-heavy and irregular. People know people who work at companies that make products bought by other people. Modeling that in a graph has three practical payoffs:

• Relationships are data, not joins. The connection between two entities is a first-class object you can store, query, and attach properties to.
• The schema can grow. Adding a new relationship type (mentors, acquired) usually requires no migration of existing data.
• Multi-step questions read like the question. "Friends of friends who like jazz" is a 2-hop traversal, not a triple self-join.

Where You've Already Seen One

• Google's Knowledge Graph powers the info panels beside search results — billions of entities about people, places, and things.
• Wikidata is an open, community-maintained KG with 100M+ items.
• LinkedIn / Facebook social graphs model people, the companies they work for, and how they connect.