Unify and manage your data

Configure FERN data selection

Learn how to limit FERN matching to a selected profile population.

FERN-based matching evaluates profile pairs within the population selected for an entity type. Define a filter to limit FERN evaluation to profiles that satisfy customer-defined business criteria.

The following roles can configure FERN data selection:

For role definitions, see About roles.

Prerequisites

Before you begin, confirm the following requirements:

  • You must have the ROLE_ADMIN_TENANT role, or a custom role that grants UPDATE on MDM:config.businessModel and READ on CatalogService.

  • You must have a pretrained FERN model available for the entity type.

  • You must have a filter expression that identifies the profiles FERN evaluates.

To configure FERN data selection
  1. From Console, select Data Modeler.
  2. In the left navigation pane, select Entity types.
  3. Select the entity type for which you want to configure FERN data selection.
  4. Select the Match tab.
  5. In the AI-Powered section, select the pretrained FERN model and open it for editing.
  6. In the Choose the dataset criteria section, select one of the two options to specify which profiles FERN evaluates.
    1. Select All profiles to evaluate every profile for the entity type.
    2. Select Define filters to restrict FERN evaluation, and then enter the filter in the Filter expression field. For example, this filter expression limits FERN to Individual profiles with an address in the city of Chicago:
      (equals(type,configuration/entityTypes/Individual) and equals(attributes.Address.City,Chicago))
  7. Select SAVE.
  8. Run the Rebuild Match Table task for the entity type to apply the change to your data.

Result

FERN evaluates a candidate pair only when both profiles satisfy the data selection filter.

Verification steps

To confirm FERN data selection, perform the following steps:

  1. In the Hub, open a profile and select the Potential Matches perspective.

  2. Review a candidate pair in which both profiles satisfy the configured filter.

  3. In the pair's matching details, confirm that AI Powered is displayed.

  4. Review a candidate pair in which either profile does not satisfy the filter.

  5. Confirm that AI Powered isn't displayed for that pair.