Use when structured fields extracted from text, OCR, or documents need Jev checks for semantic contradictions and regression-case selection.
