Skip to content
Demo
AI encyclopedia / Knowledge

What is a knowledge graph?

A knowledge graph represents things and the relationships between them. Instead of keeping a supplier, contract and department as isolated records, it can show how they connect.

By Exemplary AI · Updated · Illustrated explainer

The key idea

A graph makes relationships explicit. Each connection still needs a trustworthy source.

The idea, illustrated.

The relationship is part of the information

Figure 08
SupplierHarbour Supply
is a party to ↓
ContractC-18 · Supply agreement
owned by ↓
DepartmentOperations
replaces ↓
Previous contractC-09 · Archived
A useful connection records its source. Matching names alone does not prove that two records describe the same organisation.
Invented organisations and records. Read each arrow as a sentence: “Harbour Supply is a party to Contract C-18.” Labels explain the relationship, not just that a link exists.
  1. 01

    Identify the things

    Create distinct entities for the supplier, contract and department, using stable identifiers where possible.

  2. 02

    Name the relationships

    Record how they connect: a supplier is a party to a contract; a department owns it.

  3. 03

    Follow the connections

    Use the relationships to understand context, while checking where each fact came from and when it was last confirmed.

How a graph represents a fact

A simple relationship can be read as subject, relationship and object: “Contract C-18 — owned by — Operations.” W3C’s RDF primer describes this triple structure as a way to represent linked information. Knowledge graphs can use different technologies; they do not all have to use RDF.

The label on the relationship matters. A line between two names does not tell you whether one owns, supplies, approves or merely mentions the other. Clear relationship names turn a picture into information that can be examined and used.

Primary source: W3C: RDF 1.1 Primer

The difficult part is knowing what is the same thing

One file may say “Harbour Supply,” another “Harbour Supply Ltd.” A graph becomes useful when the system can connect records for the same organisation without merging unrelated companies with similar names. A shared identifier is stronger evidence than a matching name alone.

Relationships also change. A contract can move to a new owner, expire or be replaced. The graph needs a way to keep source and time in view. A visually impressive network is not valuable if a reader cannot tell whether a connection is current, inferred or confirmed.

An everyday example

Finding who is affected by a supplier change

A team starts from a supplier and follows links to its contracts, then to the departments that own those contracts. This makes the affected records easier to inspect together.

The graph helps assemble the context. It does not decide whether a supplier change is acceptable; the contract terms and responsible people still determine that.

A common misconception

“A line in the graph proves a relationship is true.”

A connection may come from an imported record or an extraction that needs review. Source information and identity checks matter as much as the drawing.

How Exemplary helps

See connected context in Exemplary.

Exemplary’s Knowledge Graph gives teams a visual way to explore organisation context. We help connect the relevant information so relationships can be inspected alongside the work they support.

Explore Knowledge Graph

Sources & further reading

Primary publications and documentation. Accessed 22 September 2026.

  1. 01
    RDF 1.1 Primer

    W3C · Working Group Note · 2014

Connected reading

Follow the idea further.

Understand the concept, explore the evidence or see it applied to a task.

We’re here to help

Bring us the task. Leave the AI setup to us.

Tell us what takes too much time today. We’ll help choose the right approach, set up the AI and test it with your team. You decide what a good result looks like. We help with the technology to get there.

Talk to our team