Open data / 29 published cases
Audit the classifications. Reuse the structure.
The published case library is available as structured JSON and normalized CSV tables. Claim citations, actor-scoped outcomes, source roles, review dates, evidence labels, and state-nexus labels travel with the data.
cases.json ↗Complete nested case records, including transfer chains, timelines, actions, and sources. cases.csv ↗One summary row per case. Actor names and outcomes are intentionally kept out of this case-wide table. actors.csv ↗One row per actor or affected institution, with role, actor-specific outcome, and source IDs. claims.csv ↗One row per factual or analytic claim, including stage evidence states, qualifiers, and source IDs. sources.csv ↗One row per source with publisher, dates, URL, record role, locator, archive link, and note.
Field principles
- caseId is stable across revisions.
- status describes the evidentiary or legal posture of conduct.
- stateNexus independently describes the public evidence connecting that conduct to the PRC state.
- lastReviewed and nextReview expose currency.
- Nested claims in JSON and normalized claims in CSV carry the source IDs assigned to them.
- Actor outcomes remain actor-scoped and must not be joined to another actor or count.
Use and limitations
The data summarizes public records and editorial analysis; it is not a sanctions, screening, or watch list. Inclusion must not be used as a proxy for ethnicity or nationality. Read the underlying case and sources before relying on a field.