Which confidence score does not yield an overall score at the document level?

Extraction Confidence represents how reliably an automation process can retrieve specific pieces of information from a document. While it provides valuable insights into the quality of the extracted data, this score is typically associated with individual data elements rather than providing an aggregate measure at the overall document level.

In contrast, OCR Confidence reflects the accuracy of text recognition from scanned documents, Classification Confidence indicates the likelihood that a document has been correctly categorized, and General Accuracy combines various performance metrics into a single score that assesses the overall reliability of the entire document processing task. Hence, Extraction Confidence stands out as it is primarily focused on specific elements rather than providing a broader, document-wide score.