ConverSight Web - Version 2.0.37.5

We are excited to announce the release of ConverSight web version 2.0.37.5, packed with a range of new features, enhancements and bug fixes. Here are the key highlights of this release:

New Features

Athena

  • GPT Entity Clarification: Instead of Athena randomly selecting an entity from available columns when multiple options exist, users can now specify from which table or column the entity should be selected.
  • Improved Graph Loading Performance: To expedite graph loading, a new feature has been implemented to initiate knowledge base loading upon user login or dataset switching, thereby reducing delays in graph loading times.
  • Support for Compare-Based Questions: New rules have been introduced to facilitate compare-based questions involving two different metrics.

Insights

  • Venn Chart: Implementation of Venn Diagram for data representation.

Features in UI

  • User Flow : Implemented userflow in conversight for better understanding of the application flow.

AI Workbench

  • Logs in Flow: Any skipped logs during the execution of a flow are now accurately recorded within the flow’s run status.
  • New screen in CS App: Introduction of a new screen during CS App initiation when clusters are initializing.
  • Trigger for Notifying Users of Stuck Flow Run Status: A trigger has been implemented to notify users when a flow run status becomes stuck.

Enhancements

Athena

  • Expanded Query Capabilities: Users now have increased flexibility in performing flag-based and entity count queries with additional variations.
  • Enhanced Period-Based Question Support: Added support for queries specifying periods such as “Jan-2024” or “q1-2022.”
  • Versus Query Handling: Queries containing “versus” or “vs” are now treated equivalently to compare questions for improved query interpretation.
  • Resolution of Smart Column Issues: Previously encountered issues with certain smart columns in Athena, attributed to complex joins, have been resolved. Users will no longer encounter errors such as “maximum recursion depth exceeded while calling a Python object.”

Storyboard

  • Period Selection in Storyboard Creation: Users can now select the report period when creating a new storyboard.

Bug Fixes

AI Workbench

  • Time zone fixed in Flow Run Status: Resolved issue where Flow Start time and End time in Flow Run Status were displayed in an incorrect timezone.
  • Improper temination of Scheduled Cluster: Fixed bug causing improper termination of scheduled clusters.
  • Warning logs in Smart Analytis: Addressed warning logs appearing during the creation of smart analytics.
  • New Logger component in Task and CS App: The integration of a new logger component for Tasks and CS App now organizes logs by date, allowing users to click on specific dates to view the corresponding logs.
  • Cluster initiation logs: Cluster initiation logs will now indicate whether the cluster was initiated by the user, Athena or a scheduled flow.

Athena

  • Fixed period to period queries: Resolved issue with period-to-period queries, specifically when querying for metrics spanning the current year and the previous year.

Insights

  • Unexpected Behavior on CSGrid Filter Screen: Previously, when utilizing CSGrid and attempting to filter, pressing the ESC key would remove the filter popup. However, instead of only closing the popup, the entire Explorer was being closed. This issue has been addressed.
  • Incorrect Aggregation in GEO Charts: Previously, GEO charts were defaulting to summing for aggregation. With this fix, the aggregation now utilizes the SME. If the metric is AVG or SUM, it will be used accordingly.

Storyboard

  • Error Message Alignment in Storyboard and Chat: Alignment issues with error messages in storyboard and chat interfaces have been fixed.

SME Coaching

  • Persistent Loading Spinner for Lengthy Queries: Fixed an issue where the validate button kept spinning indefinitely for queries taking more than 30 seconds to execute, without indicating failure.
  • Incorrect Publish Date Display: Even after making changes, publishing and republishing datasets, the publish date continued to display incorrectly as February. This issue has been corrected.