Definition
A product knowledge graph is a structured representation of products and their relationships — attributes, categories, variants, compatibility, and links to brands and standards — that lets systems reason over products rather than just match keywords.
Key points
- It models entities and the relationships between them.
- It powers semantic search, recommendations, and agent reasoning.
- It is built from clean, structured, normalized product data.
- It complements vectorized data and structured markup.
What does it enable?
Because it captures how products relate, a knowledge graph can answer questions a flat catalog can't — what fits this model, what replaces a discontinued part, which items share a compatible spec. That relational reasoning is increasingly what AI discovery and shopping agents depend on.
Common pitfalls
- Building a graph on inconsistent or incomplete source data.
- Modeling relationships that the underlying attributes can't support.
- Treating it as a search index rather than a reasoning layer.
FAQ
How is a knowledge graph different from a catalog?
A catalog is a list of product records; a knowledge graph adds the relationships between them, so systems can answer relational questions rather than just retrieve rows.
Why do product knowledge graphs matter for AI?
They give AI a structured, connected view of products — improving semantic search, recommendations, and reasoning. A graph is only as good as the clean data it's built from.