Skip to content

Define Fraud Decision Intelligence ontology and SHACL surface #111

Description

@mdheller

Purpose

Define the ontology and SHACL surface for the Fraud Decision Intelligence capability family.

Upstream doctrine anchor:

  • SocioProphet/ProCybernetica:docs/capabilities/fraud-decision-intelligence-control-plane.md

Minimum vocabulary

Define canonical classes/properties for:

  • FraudEvent;
  • FraudEntity;
  • FraudRelationship;
  • FraudEvidenceBundle;
  • FraudDecisionReceipt;
  • FraudLabelProvenance;
  • FraudFeatureHealth;
  • FraudScenarioCandidate;
  • FraudControlAction;
  • FraudOutcome;
  • ResidualFraudLift;
  • FalsePositiveCluster;
  • FraudRingHypothesis.

Required distinctions

The ontology should distinguish:

  • risk signal vs proof;
  • scenario candidate vs confirmed typology;
  • graph relationship vs attribution/intent;
  • observed fraud vs residual fraud;
  • weak label vs confirmed label;
  • public-synthetic evidence vs private customer evidence;
  • model recommendation vs policy-authorized control action.

Acceptance criteria

  • RDF/OWL vocabulary lands under the existing ontology structure.
  • SHACL shapes reject model-score-as-proof patterns.
  • SHACL shapes require label provenance for confirmed fraud labels.
  • SHACL shapes distinguish recommendation from authorized action.
  • Public-synthetic examples do not include private customer data.
  • Documentation states Ontogenesis owns ontology semantics, not runtime scoring or enforcement.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions