# Provenance — IOCE Core Ontology v0.1

Version 0.1, prepared 2026-09-27. The machine-readable records are
[`/provenance/sources-v0.1.json`](/provenance/sources-v0.1.json) and
[`/provenance/provenance-v0.1.json`](/provenance/provenance-v0.1.json).

## What IOCE created
- The concepts and definitions on the public reference pages: *Isocracy: The Model* (`/isocracy.html`),
  *The Rudiments of OCDI* (`/ocdi.html`), *Definitions, Methodology & Datasets* (`/methodology.html`), and the
  *Voting Influence Corruption* report (`/votes/`, including the OCDI-Gov definition).
- The OCDI formula and the OCDI-Gov adaptation, the six isocratic principles, the five branches, the provenance-tier and
  confidence scheme, the structural-heuristic scoring method, and the map layer, glyph and triad vocabularies.
- The ontology **structure** in this package: identifiers, hierarchy, relation types, status labels and these provenance
  records. This structure was created on 2026-09-27 for machine readability.

## What was derived (status `inferred`)
Items the ontology states but IOCE does not state explicitly. Each is labelled `inferred` and never presented as fact:
- a named *per-actor ownership-control divergence* (the formula's "mismatch" step);
- the **range of OCDI as published, [0, 2]**, as opposed to OCDI-Gov's documented [0, 1];
- that a resource's `institutionId` associates it with an institution (the kind of association is not stated);
- three concept-to-concept `relatedTo`/`derivedFrom` links, each with the quote that motivates it.

## What is proposed (status `proposed`)
Machine-layer structures that are **not IOCE claims**: the epistemic-status vocabulary, `SourceDocument`,
`ProvenanceRecord`, `Assertion`, the annotation properties (`quote`, `formula`, `rangeNote`, …), and the generic relations
`derivedFrom`, `hasEvidence` and `relatedTo`.

## What is unknown (status `unknown`)
Labels that appear publicly without a public definition: *stewardship score*; the four **triads** (Ownership/Control/Benefit,
Ecology/Equity/Care, Truth/Goodness/Beauty, Idea/Symbol/Object); three ownership/control-class data values; the intermediate
governance classes; the numeric OCDI proxy per class; and the "critical threshold" behind the Critical Divergence glyph.

## What came from external sources
| Source | Used for | Licence (as verified) |
|---|---|---|
| GLEIF LEI records | live Baltic ingestion | CC0 1.0 — verified 2026-09-27 |
| Latvia Register of Enterprises (3 datasets, data.gov.lv) | live ingestion | CC0-1.0 — verified 2026-09-27 |
| Estonian e-Business Register open data | live ingestion | **licence not verified** |
| Open Ownership register | live ingestion | **licence not verified** |
| Wikipedia list articles (5), retrieved 2026-06-14 | curated one-time snapshots and curated CSV | Wikipedia text CC BY-SA 4.0; status of extracted facts **needs legal review** |
| Seeded Baltic institutions (~290) | curated seed layer | **provenance unknown** |
| FEC, GovInfo, congress-legislators, Voteview, DIME | Voting Influence Corruption report | as stated on `/votes/provenance.html` (not re-verified) |
| OpenStreetMap | map tiles | ODbL 1.0 — verified |
| DW-NOMINATE (Poole–Rosenthal, Voteview) | cited measure | `externally-sourced`; not verified |

**None of these datasets is copied into the ontology.** In the live API, 326 of 503 map points carry no `source` field; their
provenance is recorded as unknown.

## Which APIs are authoritative
- For **concept definitions**, the authoritative sources are the IOCE reference pages listed above. The ontology quotes
  them, and if the two ever disagree, the page wins and the ontology has an error. Please report it.
- For **entity records and scores**, the authority is the existing read-only IOCE API (`https://ioce.alkhemy.com/api/`),
  e.g. `/api/institutions`, `/api/ocdi/map-data`, `/api/layers`. Each record's `dataTier`, `confidence` and `source` fields
  are the per-record provenance. Being able to query an endpoint does not mean you may redistribute its contents.
- For **third-party facts**, the authority is the upstream source, not IOCE.

## How provenance was established
- **Verbatim quotes**: every `documented` definition or evidence quote was checked automatically, after whitespace and
  quote-mark normalisation, against the text served at its `sourceURL` on the retrieval date. The retrieval time and the
  SHA-256 of the retrieved bytes are recorded in `sources-v0.1.json` (`retrieval`).
- **Image transcription**: the OCDI formula is published as an image (`/ocdi-formula.webp`). It was transcribed by visual
  inspection, which cannot be machine-verified. The image says "agent i" where the page text says "actor i".
- **API structure**: field names and vocabulary values were observed in GET responses. No records were copied.

Where provenance could not be established, the record says `provenance unknown`.