Learn from a Liveboard

Learning from Liveboards helps Spotter learn reusable context from selected Liveboards and apply that context when answering questions on related data models.

To enable this feature, contact ThoughtSpot Support.

Only users with data model access can use this feature.

Using Learn from Liveboard

To optimize your data model using Learn from Liveboard, follow these steps:

  1. In the Data workspace, click the name of the Model you want to optimize for Spotter.

  2. In the Model, click the More menu Chart configuration icon image, and select Enable Spotter.

  3. Click the Spotter optimization tab.

  4. Under Learn from liveboard, click Learn.

    Learn from liveboard

  5. In the selection pop-up, choose associated for dependent Liveboards and click Learn. We recommend beginning by selecting one or two Liveboards.

    Select Liveboards to learn from.

  6. The background generation should complete in five to ten minutes. During processing, a loader appears while memory generation runs in the background.

  7. Open View memory to review the generated memory and learned context.

    Click View memory and review the generated memory.

  8. Open memory summary to review related questions, then validate behavior in Spotter conversations on the same Model.

    View memory summary to see related questions.

Expected behavior

After memory is generated, Spotter may fetch this learned context during conversation and produce responses that reflect this additional context.

Be sure to validate the context by asking representative questions in Spotter to compare answer quality before and after learning.

Troubleshooting

If Learn from Liveboard is not working, check the following requirements:

  • Confirm the user has data model edit access.

  • Run learning again with one or two Liveboards, and wait the full five- to ten-minute window.

  • Check whether memory appears in View memory.

  • Validate in conversation with questions related to the selected Liveboards.

  • Review the Thinking section to verify whther additional context is being fetched while generating a response.

Limitations

  • You cannot currently edit generated memory.

  • You cannot currently modify or delete individual learned rules, you can only delete the full memory object.

  • Learning can apply across associated data models that selected Liveboards are built on.

  • Processing is asynchronous and can take at least five to ten minutes.

Was this page helpful?
×