Verified Questions for Conversational Analytics
Successful Natural Language Query (NLQ) and SQL interactions within an AI Space and semantic model can be verified. You can mark an interaction as a Verified Question by liking it, indicating that the generated SQL accurately reflects your analytical intent.
Verified questions serve as a personalization and learning input for the system, helping ensure more consistent SQL generation for similar queries in the future. They also provide trusted contextual examples that enhance the quality of conversational analytics. These verified questions are stored as your preferences and can be viewed and managed from the Manage Feedback screen, along with existing column- and value-level feedback.
Important
The user interaction for Kyvos Dialogs is the same in both the Web and Desktop interfaces. However, the screenshots for the Kyvos Dialogs Desktop and Web applications may differ. For demonstration purposes, the Web-based user interface is used to illustrate the functionality of Kyvos Dialogs.To verify the questions, perform the following steps.
Run a Natural Language Query (NLQ) in the Kyvos Dialogs to generate the response with the corresponding visualization or result.
Review the generated answer to ensure that the SQL and results correctly represent the analytical intent.
Click the Verify Answer icon displayed below the response. The Verify This Answer dialog box appears.
In the User Query field, review the query title that generated the response.
(Optional) In the Additional Context field, add any assumptions, clarifications, or supporting details related to the answer.
Click Verify to mark the interaction as a verified question.
Once verified, the interaction is stored as a trusted example and display on the Manage User Feedback page. From that page, you can approve, ignore, or delete the query.