What is a taxonomy?
A taxonomy is a controlled, tree-shaped set of categories used to classify things. Each item is assigned to one or more categories, and each category has a parent, so commercial litigation sits under litigation. The categories have agreed names and definitions and are maintained deliberately rather than growing as people invent labels. Taxonomies appear throughout a business: the product hierarchy in the ERP, the chart of accounts, ticket categories in a help desk, and classification labels on sensitive data. A taxonomy is simpler than an ontology, which also defines the relationships between concepts.
Why a taxonomy matters
Consistent categories make counting, routing, and searching possible. A law firm that lets each partner describe matters in free text cannot report how much litigation it did last year. One with a matter taxonomy can, and can train a text classification model to assign categories to new intake forms. Taxonomies also give AI agents a safe vocabulary. A support agent that must choose one of forty defined ticket categories behaves far more predictably than one asked to invent a label, and its choice can be validated against the list before it reaches the ticketing system.
What to watch for
- Keep the tree shallow, since users will not navigate past three or four levels and classifiers lose accuracy on rare categories.
- Give the taxonomy an owner and a change process, because uncontrolled additions turn it back into free text.
- Store the taxonomy as data with stable identifiers, so renaming a category does not break historical reporting.