Traceable Extraction

A technical concept in the field of intelligent document extraction, referring to the ability of extraction results to be precisely traced back to specific locations in the source document, such as page numbers, coordinates, or text spans. It stands in contrast to hallucinated outputs and is a core capability requirement in compliance and audit scenarios.