Curated list of bold-label fields that are pure process / audit metadata — never genuine journalistic content even when they appear inside a paragraph that also carries fact-bearing labels. Stripped line-by-line by stripProcessMetaLines, so per-document identification cards (Title, Beteckning, Department, Committee, Minister, Response deadline, Filed, Effective date, …) survive intact while the workflow noise (Author, Date, Confidence, DIW Score, Admiralty, …) is removed.
This is a strict subset of ADMIN_FIELD_NAMES so we never strip a label that the paragraph-level stripper considers "fact-bearing".
Added 2026-04-27 (Round 7) after the journalist-perspective audit on the latest articles found that the paragraph-level stripper was over-conservative with mixed paragraphs — per-document **Author** / **Date** / **Confidence** / **DIW Score** / **Admiralty Code** / **Self-audit cycle** / **Standard** / **Framework** lines survived inside the per-document identification cards.
Curated list of bold-label fields that are pure process / audit metadata — never genuine journalistic content even when they appear inside a paragraph that also carries fact-bearing labels. Stripped line-by-line by stripProcessMetaLines, so per-document identification cards (Title, Beteckning, Department, Committee, Minister, Response deadline, Filed, Effective date, …) survive intact while the workflow noise (Author, Date, Confidence, DIW Score, Admiralty, …) is removed.
This is a strict subset of
ADMIN_FIELD_NAMESso we never strip a label that the paragraph-level stripper considers "fact-bearing".Added 2026-04-27 (Round 7) after the journalist-perspective audit on the latest articles found that the paragraph-level stripper was over-conservative with mixed paragraphs — per-document
**Author** / **Date** / **Confidence** / **DIW Score** / **Admiralty Code** / **Self-audit cycle** / **Standard** / **Framework**lines survived inside the per-document identification cards.