0002 Following international standards: one form inside, standards outside

What

• Inside, the ledger (manifest) of a record uses only one form: common fields + a few fields for each data type (recording, scan, photo, video, text).
• When data goes out, we generate it in the international standard form that fits its type. These forms are not stored.
  • Full download: BagIt (RFC 8493)
  • Viewing and exchanging old documents, recordings and video: IIIF Presentation 3
  • Transcriptions, corrections, comments: W3C Web Annotation
  • Origin and AI labeling: C2PA + IPTC digitalSourceType
  • Scope and use: W3C ODRL 2.2
  • Lineage: W3C PROV-O
  • Names of preservation events: PREMIS 3
• Field names and values use the standard vocabularies as they are.
• We state the grade (conforms, aligned, referenced) honestly. We do not say “conforms” before checking with a validation tool.

The detailed tables are in 03-manifest sections 4 and 7.

Why

• Principle 1 and HGP-0003 article 6: a depositor must be able to leave with all of their records at any time. To be readable elsewhere, they must be in standard form.
• This connects directly with institutions such as the Korean Studies Advancement Center, with other archives and with viewers. arkive already exporting IIIF is an example.
• If we store several copies in different standard forms, they drift apart. If we generate them from one ledger, they cannot drift.
• The September “External Standards Mapping” had already covered origin and rights (C2PA, IPTC, ODRL, PROV). This time we added the type-specific forms and packaging (IIIF, BagIt, Web Annotation, PREMIS).

Other options

Option	Why not chosen
Use each type’s own standard directly as the ledger (PBCore for recordings, METS for old documents, and so on)	The ledger would take five forms, and common fields such as fingerprint, scope and consent would sit in a different place for each type. The app code would get heavy
Do everything with one C2PA manifest	C2PA is a standard for the origin of a file, so it cannot hold scope, consent, or the structure of faces and volumes
Use only our own form, and standards later	Every time data goes out it would need separate adjustment. Matching field names to standards from the start is cheapest

Where it has effect

• arkive: it already exports IIIF. Attach the original fingerprint to canvases and the volume fingerprint to the manifest, and make downloads BagIt.
• secretary: make “Download all my data” (a v0.1 gate) a BagIt package. Fields source_type and action in the manifest.

Cost

0 won. All are open standards. The validation tools (BagIt, IIIF, C2PA) are also free and public.