0021 Name file v2: ledger fields at the public address, a definitions file, a GEDCOM profile
The Korean version is authoritative. This page was translated by AI (Claude) and has not yet been checked by a person. 한국어 원문
This is the decision record for 08-Name file v2. 03-manifest was approved (2026-10-20), so this keeps the promise “v2 after 03 is approved”. It also publishes the “definitions page and mapping to other standards” from items 2 and 3 of decision 0020 (ontology).
https://hgp.hahoe.io/ns/v2.jsonld(JSON-LD 1.1 context). It keeps every v1 field with the same identifier, and adds the fields from the ledger (03), the deposit agreement entry (07), product provenance (0016, 0017) and the new relation network words (06 v0.7), making 77 fields in all.https://hgp.hahoe.io/ns/terms-v2.jsonld. Definitions of 153 fields and values (fields have Korean and English names, some values Korean only) and their mapping to other standards (Wikidata, CIDOC-CRM, PROV-O, FOAF, schema.org, Dublin Core, ODRL, IPTC). Mappings are written in SKOS:exactMatchmeans the same,closeMatchmeans similar.- What changes: values of
propandscopeare expanded to identifiers (IRIs). When the same word means different things in different fields, it is expanded separately with a per-field context (confirmedinlevelandconfirmedin a claim).source_typegoes to the official IPTC addresses.profilegoes todct:conformsTo. - Fields left out:
title,people,voices,place,sent_to, and the app processing status. They are fields that are never published, so they are not in the name file. A processor drops them, so even if they slip in by mistake, they do not go out. - v1 stays as it is. A version, once published, is never changed.
- Four genealogy words (from the jokbo tool’s sample conversion of a genealogy, 06-Relation network v0.7):
buried_atburial place,age_at_deathage at death withreckoning(Korean age reckoning (se-neun nai; one year at birth, plus one each new year) or full years of age),langto tell apart the script of a value (original Chinese characters, Hangul reading, romanization), andmarkfor a later mark on the original. - HGP GEDCOM profile gedcom-v1 (08-Name file section 7): public addresses for the extension tags
_KLUNAR(Korean lunar calendar),_KAGE(Korean age reckoning),_HGPSRC(who produced the reading),_HGPSTAT(confirmation status),_SHA256(original fingerprint) and_HGPKEY(source position). They replace the temporary names and private repository addresses that jokbo used.
Documents updated: 08-Name file v2 (sections 1, 2, 4, 6 and 7), 06-Relation network v0.7, 05-Glossary. The generator is in the site tools folder (not sent to the public zone).
- For other institutions’ software to read our ledger, the field names need a public meaning. v1 had only the fingerprint and relation network fields.
- Korean institutions cannot take our shape as it is (ontology survey). So the definitions file states, field by field, what is the same as the standards they know. Then a machine can do the conversion.
- Leaving fields out of the name file is also a form of protection. A field that holds people’s names does not go out as JSON-LD if it is not in the name file.
Tests (2026-10-20, jsonld.js 8.3.3, safe mode)
Section titled “Tests (2026-10-20, jsonld.js 8.3.3, safe mode)”A claim entry, a node entry, an external identifier claim, a ledger entry, an agreement entry, a product provenance entry, a publication entry and a fixity bundle were expanded and converted to RDF. All passed. v1 and v2 give the same field the same identifier. When used together with the IIIF context, the canvas’s hgp:sha256 is kept, and the profile in the manifest’s seeAlso keeps its IIIF meaning. terms-v2.jsonld also passes in safe mode (701 RDF lines). A claim with lang also passes. The test with the official IIIF validator will be run again when arkive switches to v2.
Other options
Section titled “Other options”| Option | Why not chosen |
|---|---|
| Define it as an OWL ontology | Automatic inference does not fit the principle of human confirmation, and no Korean institution would read it (decision 0020). SKOS mappings are enough |
Keep prop values as plain text (as in v1) |
A machine could not know that a courtesy name (ja) is Wikidata P1782 |
| Put in every field | It would open a way for fields with people’s names to go out as JSON-LD |
A global @vocab |
Every unknown field would become an hgp identifier, which gets confusing when mixed with IIIF. A per-field context is used only where needed |
Where it has effect
Section titled “Where it has effect”- arkive: can switch
hgp_contextto v2 (now v1). Runs the IIIF validator again. Uses theterms-v2mappings in the relation network export to make a flat version for Korean institutions. - secretary, provenance: use v2 when publishing the ledger as JSON-LD. Fields whose value is an object are wrapped with
@json. - jokbo: changes the
SCHMAaddresses in the GEDCOM header to the addresses in section 7. The lunar calendar is the_KLUNARcalendar +PHRASE+SDATE. - Rule table:
relations-v1.jsonstays as it is. A rule table v2 that checks the new words is published separately.
0 won.