EM-XCT-03 · Shared contract · W0
Provenance, trust and assertions
A common form for source, derivation, confirmation level, uncertainty and knowledge correction.
Partially published; research continues
Claude: independent-study-and-five-static-audits-complete-final-accept-with-limits; Grok: independent-study-and-five-static-audits-complete-final-accept-with-limits; intermediate-truncation-preserved.
Published results
Research note, in Russian: Authenticated source/agent attestations and connectors; independent-source proof; calibrated numeric confidence; sector/legal/forensic profiles and full parent holds; fine-grained disclosure and retention/erasure; durable concurrency, capacity/rollover and existing-Dimension transactional migration. This companion is a bounded trusted-host reference, not complete EM-XCT-03.
Subject boundary and candidate types
- ProvenanceRecord
- EvidenceLink
- ConfidenceAssessment
Deep research questions
- What was observed directly?
- How to mark AI synthesis versus a source-asserted fact?
- How to retract an unreliable assertion?
Verifiable invariants
- A citation does not confirm truth
- Confidence is not a probability without a methodology
- A correction preserves the previous record
End-to-end acceptance scenario
Distinguish observed, source-asserted, inference and proposal on one example and trace a correction.
Negative case
AI analysis of a local file is presented as a check of a live system.
Approaches to compare
- Current Vercy protocol/MMAS/whole-object and the published WM-XCT
- W3C PROV/SKOS/SHACL/ODRL where applicable
- Practice of source connectors, versioned schemas and policy enforcement
Candidates in the live catalogue
- WM-XCT-012 · Provenance · 0.3.0-research.1 · installable
Semantic fit requires boundary research; a published model does not by itself complete this card. - WM-XCT-026 · Quality / Confidence · 0.3.0-research.1 · installable
Semantic fit requires boundary research; a published model does not by itself complete this card. - WM-XCT-028 · Evidence / Rationale · 0.3.0-research.1 · installable
Semantic fit requires boundary research; a published model does not by itself complete this card.
Result requirements
Every card is executed together with the full research contract: definitions, fields and cardinalities, lifecycle, sources, data mastership, rights, the five object facets, at least eight invariants, positive and negative examples, dependencies, migration and applicability limits.