Skip to content
LiveNext 10:46:09

Definition

What is Knowledge graph?

Knowledge graph is a knowledge graph is a database that stores facts as connected entities and relationships, such as products, brands and prices, so software can look up and reason over how things relate.

A knowledge graph stores information as a network instead of a table. Each thing, such as a person, a company or a product, is a node. Each link between things, such as "made by" or "costs" or "is a version of," is an edge. Together they form a map of facts that software can walk through.

How it works

The basic unit is a simple statement with three parts: a subject, a relationship and an object. "Running shoe X, made by Brand Y" is one. "Brand Y, based in Oregon" is another. Because both mention Brand Y, they connect, and a program can answer a question that spans them, like "which shoes come from Oregon brands?"

A table would need you to plan that join in advance. A graph makes new connections cheap to add and to query, which is why it suits messy, fast-growing data.

Building one takes work. Facts must be collected from many sources, matched when the same thing appears under different names, and kept current. Much of the effort goes into that cleanup.

Why it matters to you

Search engines use knowledge graphs to power the fact boxes you see for people, places and products. Online stores use them to describe what a product is, so listings from different sellers can be compared.

For AI, a graph gives a model something firmer than memory. A model can ground its answer in graph facts, which helps reduce hallucination. Graph data can also be fed into retrieval-augmented generation so the model looks up related facts before it answers.

Picture a city map. A list of street names tells you what exists. The map tells you what connects to what, and that is the part you need when you are trying to get somewhere.

A vector database finds items that are similar in meaning by comparing embeddings. A knowledge graph finds items that are explicitly linked. They solve different problems and are often used together: vectors to find likely matches, the graph to confirm how they relate.

For AI shopping agents, product data is the catalog they read. A graph that cleanly lists what each product is, who sells it and at what price is the kind of structure an agent can act on, which is why agentic commerce puts so much weight on clean product data.

Questions people ask

What is the difference between a knowledge graph and a vector database?

A knowledge graph stores explicit links between facts, while a vector database finds items that are similar in meaning by comparing embeddings. Many systems use both.