Relationship Validation Ontology

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

flow "RelationshipValidationOntology" v1.0 {
  ontology:
    id: "relationship-validation-ontology"
    parent: "ai-semantic-library-relationships"
    purpose: "Validate typed semantic relationships against ontology constraints, evidence, provenance and versioned reasoning rules."

  types:
    Entity: {id: string, ontology: string, class: string, version: string?}
    Relationship: {id: string, subject: string, predicate: string, object: string, context: string?, confidence: number, evidence: string[]}
    Constraint: {id: string, kind: enum[domain, range, cardinality, disjointness, temporal, logical, provenance, custom], expression: string, severity: enum[error, warning, information]}
    Evidence: {id: string, source: string, timestamp: datetime?, provenance: string, reliability: number?}
    ValidationResult: {relationship_id: string, status: enum[valid, invalid, indeterminate], violations: string[], warnings: string[], checked_at: datetime, rule_version: string}

  predicates:
    subject_of(entity: Entity, relation: Relationship)
    object_of(entity: Entity, relation: Relationship)
    governed_by(relation: Relationship, constraint: Constraint)
    supported_by(relation: Relationship, evidence: Evidence)
    validated_as(relation: Relationship, result: ValidationResult)
    maps_to(entity_a: Entity, entity_b: Entity)

  constraints:
    - "Every relationship subject and object must resolve to an identifiable entity or explicitly permitted literal."
    - "Predicates must resolve to declared properties with compatible domain and range."
    - "Confidence values must lie within [0,1]; absence of evidence is not evidence of validity."
    - "Enforce cardinality, disjointness, temporal and logical constraints under the selected ontology version."
    - "Do not treat an open-world absence of a triple as a contradiction."
    - "Differentiate structural validity from factual truth and evidential support."
    - "Do not infer identity from aliases, lexical similarity or a mapping alone."
    - "Retain validation provenance and all conflicting evidence."

  inference:
    rule type_check:
      if declared_domain(p,c) and relation(s,p,o) and type_of(s,t) and incompatible(t,c)
      then flag_violation(s,p,o,"domain_mismatch")
      confidence: "deterministic only under explicitly selected closed validation profile"
    rule evidence_gap:
      if relationship(r) and requires_evidence(r) and no_qualifying_evidence(r)
      then flag_warning(r,"insufficient_evidence")
      confidence: "do not assert falsehood"
    rule validation_aggregation:
      if all_required_checks_pass(r) and no_blocking_violation(r)
      then validated_as(r,"valid")
      confidence: "validity is relative to rules, data snapshot and ontology version"

  SEMANTIC_EQUIVALENCE:
    - "Use versioned canonical identifiers and explicitly scoped equivalence mappings."
    - "Keep equivalentClass, equivalentProperty, sameAs and approximateMatch semantically distinct."
    - "Revalidate when a mapping, schema or ontology version changes."

  CONFIDENCE:
    range: [0,1]
    factors: [source_reliability, evidence_quality, mapping_certainty, temporal_relevance, inference_assumptions]
    policy: "Report evidence confidence separately from binary or three-valued validation status; never manufacture certainty by chaining."

  VALIDATION:
    stages: [resolve_identifiers, check_types, check_domain_range, check_cardinality, check_logical_consistency, check_temporal_scope, inspect_evidence, record_results]
    outputs: [valid_relations, invalid_relations, indeterminate_relations, violations, warnings, audit_trail]
    policy: "Use explicit closed-world constraints where completeness is required; otherwise preserve unknown status."

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

  UPDATE_MODEL:
    strategy: "versioned incremental revalidation with provenance preservation"
    steps: [ingest, normalise, detect_changes, identify_affected_relationships, validate, reconcile, publish]
    rollback: true

  AI_CONSUMPTION:
    representation: "typed knowledge graph with machine-readable validation reports"
    outputs: [validated_triples, constraint_violations, confidence_annotations, provenance_records, change_logs]
    safeguards: [no_unsupported_identity_merges, no_closed_world_assumptions_without_declaration, no_confidence_inflation, preserve_conflicting_evidence]
}

END FLOW ALGEBRA