Watson Knowledge Catalog (WKC) - Data Governance and Quality

What's New in Watson Knowledge Catalog on Cloud Pak for Data v3.0

By Emma Tucker posted Tue April 28, 2020 01:40 PM


With the release of Watson Knowledge Catalog on Cloud Pak for Data v3.0, we’ve introduced some new capabilities to improve user experience and reduce manual efforts through machine learning.

New look and feel – All of IBM including marketing and products are standardizing on a single, open sourced style guide. This means customers with multiple IBM products will see more consistent experiences between those products and take less time to familiarize themselves with something new. It also means faster UI development time to create features our customers love and a new, sleek look.


Support for 9 new languages – Now Watson Knowledge Catalog can be translated into Brazilian Portuguese, French, German, Italian, Japanese, Russian, Simplified Chines, Traditional Chinese, and Spanish. 

Negative term classification – A user's choice to reject an automatically suggested business term is saved and used for more accurate future term suggestions. When analysis is re-run, the system now remembers a user's choice to manually reject a term assignment, and the ML model is re-trained for more accurate term assignments.  The user also sees which, if any, business terms have already been Published (to the catalog), and how the term was assigned (by ML, manually, or by data classification).

negative term classification   

Data quality score over time - We've added the ability to view trends in Data Quality score over time. A user can quickly visualize how their Data Quality has changed over time by data asset, based on their specified time interval

data quality score over time 

Column filtering for relationship and overlap analysis - Our users rely on Relationship and Overlap analysis to be their source of truth for dependencies and completeness between all of their data assets. When searching for all parts of particular records for archival, deletion, movement, or data fabrication, WKC’s primary and foreign key analysis is the key to allowing users to act with confidence. However, that degree of confidence typically comes with a cost of comparing the many parts of different data assets to find commonalities and relationships. Our new column filtering functionality allows user to only analyze the data most relevant to their use case by scoping into or filtering out columns by name, column type, or first N candidates. 

column filtering 

Data quality settings enhancement - Users can save time by navigating the product efficiently and having their preferences saved. Preferences like sorting and searching are persisted not by browser or session, but by the user's account, so important details are preserved by the product. We persist pagination settings and tile vs table view on both Projects and Data Assets views.


Automatic data class creation - Quickly create and assign data classes to clusters of similar columns. Column Similarity Analysis uses a patent-protected algorithm that captures the 'fingerprint' of columns. Column Similarity Analysis can find and create no-code Data Classes for clusters of similar columns across an entire Project. Or, it can build a cluster from a user-specified column for more targeted Data Class Creation. It’s a quick and easy way to create data classes and presents a simple method to find and govern clusters of similar columns across all your Project's Data.

Auto data class creation  

More powerful data protection rules – Now users can build data protection rules based on the Classification of an asset.  These rules can also be used when building new views of data through data virtualization.

Classification in data protection rules 

Enhanced activity panel on governance artifacts  - The activity panel captures a full history of governance artifacts (terms, policies etc...), so users can see all changes and approvals that have happened over time to the artifact. This pulls from details captured via workflow on each governance artifact including who completed the task, what they changed, and their comments.

Activity Panel

Smarter global search – While searching for an object, cp4d will recommend search strings based on what you’re typing. This makes search more similar to Google by providing auto-completion suggestions.


New data sources and connections – We’ve added connections for Impala and Planning Analytics.





Tue May 26, 2020 09:39 AM

Thanks Emma. I was referring to your presentation on AI and Governance. I did listen to Marc's presentation as well.


Tue May 26, 2020 05:56 AM

Hi Raja - If you return to the user group presentation, you'll find a recording and you should be able to download the presentation. Are you referring to mine about AI Governance or Marc's about WKC?

Mon May 25, 2020 02:30 PM

I viewed your User Group presentation. Would it be possible to get the slides from the presentation? Thanks