Filtering and searching
The Data Model Descriptions page provides toolbar options that help you work with large data models.
- Filter
- Select a filter to show enabled elements only, disabled elements only, or all elements.
- Sort
- Reorder elements alphabetically.
- Search
- Specify text to find a measure, attribute, or saved expression by name.
Best Practices
- Write specific descriptions. For example "Sales revenue before tax, in USD, for completed orders only" is far more useful than "Sales amount."
- Use labels that match how your users talk about data. If the common name is "Gross Sales," use it instead of the system name "Sum_sale_of_product."
- Enable only what is needed. Enabling too many elements reduces AI accuracy by increasing the chance of irrelevant elements being selected.
- Start with your most-used measures and attributes. Do not configure everything at once. Begin with the elements your users ask about most.
- Provide labels and descriptions even for elements you do not enable. This improves the quality of Insight Summaries, which use this context regardless of the element's enabled status.
- Test your configuration using the Prompt Playground before making AI features available to end users.
Note:
- Data Model Descriptions include the elements from the Default Subject Area. Custom Subject Areas are not supported for GenAI configuration.
- In a parent/child workspace setup, labels and descriptions must be configured in the child (consumer) workspace. They cannot be inherited from the parent workspace.