Anatomy of a Modern Semantic Model
A robust semantic architecture completely abstracts the physical database, replacing wide, denormalized tables with a logically mapped network of interrelated business concepts. This mapping is highly structured and typically managed via YAML configuration files within a version-controlled repository. A standard semantic model is composed of five primary metadata components:
Entities
Act as the structural anchors of the semantic graph, mapping logical business concepts to underlying database tables.
In practice
Identifying an order_id as the primary key of an Orders table, or defining a “Customer” dataset and adding a short description of the business meaning.
Dimensions and Attributes
Provide the qualitative context required to group, slice, filter, and segment the data.
In practice
Geographic regions, industry types, and time dimensions with native multi- grain support (day, month, year).
Semantic Layer Guide © Scalefree International GmbH 2026
11
Powered by FlippingBook