A W3C PROV-aligned graph of entities, activities, and agents supporting origin and transformation queries.
Lineage is recorded in tables shaped by the W3C PROV data model: activities that acted, agents responsible for them, and typed relations connecting entities to the activities and agents that generated, used, attributed, and derived them.
Aligning to an established model rather than inventing a proprietary schema is deliberate. Provenance that cannot be exchanged between institutions cannot support a federated verification layer.
Because lineage is structured, the questions institutions actually ask become queries rather than investigations: where did this information originate, what datasets contributed to this output, which models transformed this knowledge, and which parties interacted with this asset.
Referential integrity is enforced at the database level, so the graph cannot reference an asset or agent that does not exist.
The provenance graph is the substrate on which the risk mechanisms operate. Entropy modeling reads the bounded neighborhood around an asset; transfer-risk analysis walks derivation ancestry and cross-asset references; impact propagation traverses dependency-bearing edges.
Each therefore inherits the same guarantee: results are derived from a recorded structure that can be inspected, rather than from opaque inference.
Entities, activities, agents, and typed relations.
Enforced at the storage layer, not by convention.
Origin and transformation answered directly.
The graph the risk mechanisms traverse.