Register an AI model (in preview)

An AI Model is a statistical model that is trained on datasets in order to make predictions on new data. You can integrate AI/ML models in Collibra or manually create AI Model assets and link them to the relevant AI use cases.

Requirements and permissions

Important This feature is in preview. To register an AI model, you need to enable Model Registry. For guidance, go to Set up AI Governance.

You have:

  • A global role with the Product Rights > Assessments global permission.
  • A global role with the AI Governance > Register AI Model global permission.

The out-of-the-box AI Business User and Data Scientist global roles have these permissions.

Steps

  1. From the Model Registry page, click Register new AI model.
    The Register new AI model dialog box appears.
  2. Enter the required information.
    FieldDescription
    NameThe name of the new AI Model asset.
    Type

    The asset type of the new AI model.

    You can choose the AI Model asset type or any child asset type of AI Model, including custom child asset types of AI Model.

    Domain

    The domain in which you want to create the new AI Model asset.

    Note  If you know that a certain domain exists, but it doesn't appear in the drop-down list, it could mean that:
    • The domain is of a type that has not been configured for this asset type. For more information, go to Assign a domain type to an asset type.
    • You are subject to view permissions restrictions. By default, all users can view all assets in all communities and domains. If a community or domain is restricted by view permissions to which you are subject, the domain is not shown in the drop-down list.
    Owner

    The Collibra user or user group who will act as the owner of the AI model. They are assigned the Owner resource role for the AI Model asset.

  3. Click Register.
    A new AI Model asset is created in the specified domain.
Tip You can configure which asset status is assigned to newly registered AI Model assets, and configure the progression of statuses to define the lifecycle stages through which your AI models will evolve. Go to Configure AI Governance lifecycle stages.

What's next?

  • On the AI Model asset page, use the Lifecycle tracker features to:
    • Start an assessment
    • Add lifecycle activities, and more.
  • Edit the AI Use Case asset, as necessary.
    Important 

    Be mindful of the edits you make on the asset page before the assessments are completed and approved. When an assessment is approved, depending on the assessment type, certain assessment responses are copied to the asset page. If you manually edit the attributes of an asset, some of the values might be overwritten when the assessment is approved. For more information, go to:

    Tip If you want to manually fill out the details without conducting the prescribed assessments, you can create the asset via the global Create button. When you create an asset that way, no assessments are started. You can then add details to the various sections of the asset page as you would an asset of any other type, knowing the values won't be overwritten by an assessment. When you're on the asset page, remember to switch on the Show Empty Optional Values option, to see all of the available fields.