Table of Contents

Dimensions

Dimensions are central to a semantic model for three reasons:

  • Aggregation. Dimensions let you roll up data over structures such as organization, product and time hierarchies. Data is usually stored at a finer granularity than the level the business reports on.
  • Relationships. Dimensions form the backbone of the relational model that ties the datasets and business metrics together.
  • Metadata. Dimensions describe the datasets and business metrics with attributes such as product category, color, size, weight, market, region and customer group. This enables flexible analysis across different dimensionality.

Properties

Name Description
Source The name of the dimension table or view that the dimension references.
Description What the dimension represents. Optional, but recommended: it helps the AI agent understand the data and when to use it.

Columns

Columns define which columns of the source table or view are exposed to the AI agent. This lets the agent fetch the dimension in tabular format for analysis.

Property Description
Key column The name of a dimension column to make available to the AI agent. This includes ID columns, attribute columns and other descriptive columns.
Display column The column that holds the common name for a key column. Only applicable to level ID columns such as DepartmentID and L1_ID.

Business systems usually store only member IDs in transactions (such as department or product IDs), not names like "Engineering" or "T-Shirts". Users, however, ask for data by name, for example: "What is the total number of FTEs in Engineering?" By pairing a key column with a display column, the AI agent can translate the name into the matching key values and filter the result correctly.
Column type

Column details

Column details define the role a column plays in the semantic model: its relationships to columns in other datasets and dimensions (typically primary key / foreign key), and its place in a hierarchy.

Relationship

Name Description
Is relationship Whether the column defines a relationship to a column in another dataset or dimension. Typically enabled for key columns such as DepartmentID, AccountID and ProductID.
Relationship name The name of the relationship the column belongs to. You can choose any name. All columns that share a relationship name belong to the same logical relationship, so you define a relationship by choosing a name and reusing it across datasets and dimensions.
Role Primary, Foreign or Unique. The relational role of the column.
- Only one column in a relationship can have the Primary role.
- Dimensions usually hold the Primary columns.
- Dimension columns in datasets (DepartmentID, AccountID, etc.) usually have the Foreign role.
Note

In a dimension, enable Is relationship only for the leaf-level ID column (such as DepartmentID), and always give it the Primary role.

Hierarchy metadata

Name Description
Is hierarchy column Whether the column represents a level in a hierarchy. Usually enabled only for columns that hold the member ID at each level, such as DepartmentID (leaf level), L1_ID and L2_ID.
Level name A name that tells the AI agent what the level represents, for example "Region" or "Country". If a user asks for sales by region, the level name lets the agent roll up the data to the correct level.
Level number The position of the level in the hierarchy. Helps the AI agent understand the structure of the dimension.
Parent The parent of the level. Helps the AI agent understand the structure of the dimension.