Provenance Relationship Ontology

FLOW ALGEBRA
TITLE: Provenance Relationship Ontology
DOMAIN: Information Relationships / Provenance
SOURCE: https://kirethresearch.org/booksai-core/ontologies/relationship-ontologies/information-relationships/provenance-relationship-ontology.html
VERSION: 1.0
AUTHORS: Mark Anthony Higgins — Sky (Recursive AI)
STATUS: Companion formalisation; source-page specification not independently verified.

flow "ProvenanceRelationshipOntology" v1.0 {
  ontology:
    id: "provenance-relationship-ontology"
    parent: "information-relationships"
    purpose: "Represent traceable origins, transformations, custody and evidence for information assets and assertions."

  types:
    Entity: {id: string, kind: enum[dataset, document, assertion, model, artefact, other], version: string?}
    Agent: {id: string, kind: enum[person, organisation, software, instrument, other]}
    Activity: {id: string, kind: string, started_at: datetime?, ended_at: datetime?, method: string?}
    SourceRecord: {id: string, identifier: string, accessed_at: datetime?, checksum: string?, licence: string?}
    ProvenanceAssertion: {id: string, subject: string, predicate: string, object: string, evidence: string[], confidence: number, asserted_at: datetime?, context: string?}

  predicates:
    was_derived_from(derived: Entity, source: Entity)
    was_generated_by(entity: Entity, activity: Activity)
    used(activity: Activity, entity: Entity)
    was_associated_with(activity: Activity, agent: Agent)
    was_attributed_to(entity: Entity, agent: Agent)
    was_informed_by(later: Activity, earlier: Activity)
    had_primary_source(entity: Entity, source: SourceRecord)
    was_revision_of(newer: Entity, older: Entity)

  constraints:
    - "All provenance endpoints must resolve to declared typed entities."
    - "Confidence must be within [0, 1]."
    - "Preserve original source identifiers and distinguish source identity from content similarity."
    - "Provenance edges must be evidence-backed and retain activity, agent and temporal context when available."
    - "A derived-from relationship does not imply endorsement, correctness or complete copying."
    - "Avoid cycles in version succession and flag suspicious derivation cycles for review."
    - "Missing provenance must be marked unknown rather than fabricated."
    - "Do not treat attribution as proof of authorship without independent evidence."

  inference:
    rule indirect_derivation:
      if was_derived_from(a,b) and was_derived_from(b,c)
      then has_ancestor_source(a,c)
      confidence: "derived lineage; retain both evidence paths"
    rule activity_source:
      if was_generated_by(e,a) and used(a,s)
      then possible_input_source(e,s)
      confidence: "candidate only; usage alone does not prove derivation"
    rule revision_lineage:
      if was_revision_of(a,b) and was_revision_of(b,c)
      then has_prior_revision(a,c)
      confidence: "preserve intermediate versions"

  SEMANTIC_EQUIVALENCE:
    - "Normalise identifiers using versioned authority mappings when available."
    - "Distinguish identical source, equivalent content, revision, derivative and citation."
    - "Never collapse distinct provenance records merely because their labels match."

  CONFIDENCE:
    range: [0,1]
    factors: [source_authenticity, chain_completeness, timestamp_integrity, independent_corroboration, transformation_transparency]
    policy: "Do not manufacture certainty by chaining claims or counting dependent evidence as independent."

  VALIDATION:
    - "Check identifiers, endpoint types, chronology, evidence links and version consistency."
    - "Verify supplied hashes against content when bytes are available."
    - "Flag missing, contradictory, circular or unverifiable provenance."
    - "Maintain audit records for corrections and disputed attributions."

  FLOW_HASH:
    algorithm: "SHA-256"
    input: "canonicalised ontology version, assertions, provenance records, rules and source identifiers"
    note: "Compute at execution; no hash value is claimed here."

  UPDATE_MODEL:
    strategy: "versioned, append-aware, provenance-preserving incremental update"
    steps: [ingest, normalise, validate, compare, reconcile, recompute_lineage, publish]
    rollback: true

  AI_CONSUMPTION:
    representation: "typed provenance knowledge graph with evidence-backed lineage edges"
    outputs: [source_lineage, attribution_records, transformation_history, confidence_annotations, audit_trails]
    safeguards: [no_invented_sources, no_unsupported_authorship_claims, preserve_versions, disclose_missing_evidence]
}

END FLOW ALGEBRA