Dynamic Search Priority at Conversation Time

While you can configure default search priorities at the data source level, you can also override these priorities dynamically at runtime during a live conversation.

Info: This feature is available starting with Druid 10.1.

To adjust search priorities dynamically:

  1. Navigate to the desired flow step where you want to apply the priority change.
  2. In the step configuration editor, click on Set variables.
  3. Set [[Intent]].KBWeights using a JavaScript function that maps your data source names to the desired priority levels (low, medium, or high).
    4. Copy

    Example

    (function(){
        return {DataSource1: "low", DataSource2: "high", DataSource3: "high"};
    })()
  4. Save the step.
IMPORTANT!
  • Case Sensitivity. While search priorities (e.g., "low", "high") are case-insensitive, data source names are case-sensitive. Ensure you match the exact capitalization of your data source names.
  • Conversation Reset. Runtime weight overrides apply only to the active session. Upon conversation reset, the system automatically reverts to the default priorities configured at the data source level.