Overview

Semantics are the organizational layer that connects raw data assets to the business concepts they represent. Every major surface in Prizm — the asset catalog, quality score groupings, RBAC rules, and criticality scoring — uses semantic context to filter, group, and prioritize. Without semantics, assets are unnamed tables; with semantics, they have owners, business meaning, and a place in the organization.

The five semantic types

Type	Purpose	Applies to	Cardinality
Domain	Business area ownership and accountability	Assets, Attributes, Products, Applications	0–1 per asset / attribute; exactly 1 per product
Product	Groups assets around a business concept with defined quality expectations	Assets, Attributes	Many assets per product; exactly 1 owning domain per product
Application	Represents source systems, processing platforms, and consumption tools	Assets	Many assets per application; many domains per application
Tag	Flexible labels for filtering, grouping, and policy targeting	Assets, Attributes	0–many per asset or attribute
Term	Primary business definition for a data attribute	Attributes	Exactly 1 per attribute

How they relate

Domains establish the top-level ownership boundary — they answer the question “which business area is accountable for this data?” Each Domain can have Sub-domains for finer-grained ownership (e.g., Finance → Accounts Receivable). Products live inside Domains. A Data Product groups the assets that deliver a specific business capability (e.g., a “Customer 360” product containing customer_dim, orders_fact, and churn_score). Every product has exactly one owning domain. Applications are independent of the domain hierarchy. An Application — such as SAP, Snowflake, or Tableau — can span multiple domains and link to multiple products. It represents the system layer: where data is created, where it is processed, and where it is consumed. Tags and Terms operate at the asset and attribute level. Tags are free-form and can stack — an asset can be PII, Certified, and Production simultaneously. Terms come from the Glossary and are assigned one per attribute — the single authoritative business definition for that column.

Domain
  └── Sub-domain
        └── Product  ←── Application (many-to-many)
              └── Asset
                    └── Attribute  ←── Term (Glossary)
                                   ←── Tag(s)

Where semantics appear

Platform surface	Semantic types used
Asset catalog — filter panel	Domain, Product, Application, Tag, Term
Quality score page — grouping dimensions	Domain, Tag, Application
Criticality scoring — inputs	Domain, Product membership
RBAC — resource-level access control	Domain, Product, Application, Tag, Term
Converse — natural language queries	All five types (semantic search)
Alerts / Issues — scoping	Domain, Product, Tag

Domain

Define business areas and sub-domains. Set accountability and group assets by ownership.

Product

Group assets into data products with health metrics, SLAs, and adoption tracking.

Application

Tag

Apply flexible labels to assets and attributes for filtering and policy targeting.

Glossary

Build a business vocabulary with Glossaries, Categories, and Terms linked to attributes.

Get Started

Core Concepts

Platform

Architecture

Data Sources

Integrations

Security & Access Control

Deployment

Help

Settings

MCP

Release Notes

API Reference

The five semantic types

How they relate

Where semantics appear

Domain

Product

Application

Tag

Glossary

​The five semantic types

​How they relate

​Where semantics appear

Domain

Product

Application

Tag

Glossary

The five semantic types

How they relate

Where semantics appear