Some source documents may have defects that result in extraction accuracy that is less than 100%. For example, some documents may be based on scanned sources, utilize complex images, use unusual fonts, or have spelling errors or omissions. As a result, the extraction service includes an extraction confidence score that is then used by the automation process to determine if the extracted data should be manually reviewed. If the extraction confidence score falls below the configured threshold, an exception flow is triggered to involve manual review by a human.

Note: In this release, the confidence threshold cannot be adjusted by application editors.

Did this answer your question?