Enhanced Datasets configurations on the platforms:
Allow versioning and approval workflows in dataset registry
Enhanced registry page with additional notes, permissible purpose, and keywords
Create artifacts for datasets which can be tracked in the Artifact Repository
Allow modelling and non-modelling datasets by providing an optional target feature that can be used with the dataset
Lineage tracking for datasets
Ability to attach Flags for datasets
Improved performance for PMML based models in Model Registry
Enable copying Rules in Quality Check
Policy and Strategy Management
Simplified UX to associate models to a policy using ModelSuite
Common
Ability to share objects with other users in the Draft and Pending Approval stage. This enables more collaboration when developing objects before it gets Approved.
Approval Workflows - The previous functionality provided by Approval Committee has been renamed to Approval Workflow with enhanced capabilities added:
Veto power - to lock objects from being changed during reviews
Allow the same user to perform multiple responsibilities in a workflow
Add ability to add informational comments to an ongoing review
Enhanced timeline-view to track historical activities in an approval, which show simulations and comments
Allow sending a new job to reviewers when re-submitting the request based on changes or information requested
Allow editing the Permissible Purpose for an object in a cloned version
Allow string comparison operators like <, >, <=, >= in quality check rules and simulation filters
Enable exporting artifacts for Draft and Pending Approval objects. To simplify debugging and testing for artifacts before it is sent for approval
Technical
Enable customization for various aspects of the platform to integrate with 3rd party systems using generic methods like APIs, FTP, SSH, etc.:
Allow shortcuts to other tools in the main navigation bar
Allow exporting documentation to system of choice
Allow exporting artifacts to system of choice
Allow reviews and approvals from external tools in an automated manner
Enable users to hook into corridor’s core actions using signals - like the review-signal. This is an advanced feature.
Enable additional APIs to be added to Corridor which can be used to fetch information using the corridor python package via REST APIs
Improved Error messages for various cases like:
Permission issues when a user accesses an object
Definition related errors in simulations, comparisons, and validation
Errors in the dependent variable computation for models