Knowledge Graph

Data stored as connected facts, so an agent can follow the links to an answer.

Definition

A knowledge graph stores data as connected facts: things (nodes) linked by relationships (edges). An AI agent can follow those links to answer complex questions, such as which product treats a specific pest on a specific crop. In commerce, it usually combines product data and customer data.

Key points

<ul><li>A database row says "Product X, Category Y". A knowledge graph also says how X relates to crops, pests, customers and past orders.</li><li>Example: "insecticide for okra" is answered by moving from the crop (okra) to the pests that affect it, then to the products that target those pests.</li><li>It has two parts in commerce, the Product Knowledge Graph and the Customer Knowledge Graph, joined at the SKU. "Send me 20 MCB, 16 amp, same as last time" is resolved in one pass: the customer, their usual product, the exact variant and the distributor who fulfils it.</li><li>It gets richer with every conversation and transaction.</li></ul>