Data Governance and Cataloging
Domain 4 tests Watson Knowledge Catalog: publishing assets, governance artifacts, and enforcement.
Catalog versus project
A project is a working area. A catalog is a curated, governed collection that can be shared across projects. Publishing an asset creates a catalog entry; governance artifacts (business terms, data classes, policies, rules) attach meaning and control.
Governance workflow
Create categories and business terms, define data classes with matching criteria, import metadata, enrich assets with terms, then publish. Data protection rules use data classes or terms to mask, filter, or deny access. Many artifacts move through a workflow (draft, review, published), so a rule that is not published will not enforce.
Worked example
A column containing email addresses must be masked for non-stewards. Create a data class that matches email patterns, associate it with a business term such as “Email Address,” build a data protection rule that masks the column for users outside a steward group, publish the rule, and verify with a non-steward account. The reasoning: classification drives policy; policy must be published and tested.
Traps
Confusing catalog membership with project membership. Expecting lineage without metadata import or job instrumentation. Creating a rule but leaving it in draft. Assuming every asset is governed automatically; assets must be assigned or discovered.