Use this API to update the ground truth (correct extraction values) for a document that has been previously imported to the accuracy center.

Two ways to import documents:

Import from datalog (when datalog is enabled) - You can provide the ground truth directly during import via the "Import document from datalog" API. If ground truth is provided, the document becomes immediately active for training.
Via /v2/extract with import-to-accuracy-center header - Documents imported this way are stored but inactive. You must use this "Update ground truth" API to provide the correct values before the document can be used for Smart Learning training.

The document_data parameter must contain the correct extraction values. The required schema matches the data object structure returned by the /v2/extract endpoint.

Typical workflow for /v2/extract imports:

Call /v2/extract with import-to-accuracy-center header - document is stored but inactive
Call this API to update the ground truth - document becomes active for training
Smart Learning uses the document to improve future extraction accuracy

Note: Datalog is an enterprise feature. Please contact sales if you are interested in enabling this capability.