Data Availability
Generated by scripts/generate-traceability-report.py on 2026-06-28T10:31:47.
Project scripts, schemas, manifests, source registries, analysis artifacts, validation outputs, and public review pages are intended to be inspectable in the repository. Corpus text reuse is governed by each document’s source metadata, rights status, storage policy, and source edition or translation notes.
Raw texts that are public domain and marked as committed may be inspected from the repository. Metadata-only, local-only, restricted, or unresolved sources should be reacquired from their cited source records rather than reused from the workbench. Publication quotations should still be checked against the named edition or manuscript source before final use.
Model-reliability schemas, prompts, code, status records, and rights-safe aggregate diagnostics may be shared. Packet payloads and raw model responses inherit the underlying source’s rights and storage constraints. Artifacts that reproduce local-only, restricted, metadata-only, or unresolved source spans must remain local or be withheld; public reports should use stable identifiers, aggregate measures, provenance metadata, and short compliant spans instead.
Human-reliability schemas, case status records, aggregate reports, and correction-candidate metadata may be shared when rights-safe. Human packet payloads, raw coder submissions, adjudication submissions, and adjudication working materials inherit both source-document restrictions and coder-storage restrictions. Public reporting should use cohort IDs, stable artifact IDs, aggregate metrics, status fields, and short compliant spans rather than raw restricted payloads.
Frozen Corpus Boundary
The expanded pre-v1 corpus boundary is frozen as of 2026-06-28 at 6 cases and 41 manifest documents. Deferred candidate cases or documents from the expansion matrix are future work and are not part of the v1 reliability or claim-promotion cycle.
| case_id | manifest_documents | boundary_status |
|---|---|---|
| am-rev | 11 | frozen-pre-v1 |
| fr-rev | 2 | frozen-pre-v1 |
| hitler | 8 | frozen-pre-v1 |
| lincoln | 5 | frozen-pre-v1 |
| napoleon | 11 | frozen-pre-v1 |
| wwi-britain | 4 | frozen-pre-v1 |
Corpus Availability Snapshot
| case_id | text_id | rights_status | git_tracking | expected_local_path | known_limitations |
|---|---|---|---|---|---|
| lincoln | lincoln-lyceum-address | public-domain | committed | cases/lincoln/corpus/raw/lincoln-src-01-lyceum-address.txt | Gutenberg/Lapsley transcription is adequate for pilot annotation but should be cross-checked against Basler before publication-level quotation. |
| lincoln | lincoln-first-inaugural | public-domain | committed | cases/lincoln/corpus/raw/lincoln-src-04-first-inaugural.txt | Public-domain transcription is adequate for draft annotation; exact wording should be cross-checked against LOC or Collected Works before publication-level quotation. |
| lincoln | lincoln-special-message-1861-07-04 | public-domain | committed | cases/lincoln/corpus/raw/lincoln-src-05-special-message-1861-07-04.txt | Long presidential message increases annotation burden; exact wording should be cross-checked against LOC or Collected Works before publication-level quotation. |
| lincoln | lincoln-gettysburg-address | public-domain | committed | cases/lincoln/corpus/raw/lincoln-src-02-gettysburg-address.txt | The Bliss copy is used for pilot consistency; variant manuscript wording should be checked before claims depend on exact phrasing. |
| lincoln | lincoln-second-inaugural | public-domain | committed | cases/lincoln/corpus/raw/lincoln-src-03-second-inaugural.txt | Reliable for pilot annotation; manuscript cross-check remains required before publication-level quotation. |