Predicate Relationship Ontology

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

flow "PredicateRelationshipOntology" v1.0 {
  ontology:
    id: "predicate-relationship-ontology"
    parent: "knowledge-graph-relationships"
    purpose: "Represent RDF predicates as typed, identified relationships connecting subjects to objects, with explicit semantics and provenance."

  types:
    IRI: string
    Resource: {id: IRI}
    BlankNode: {local_id: string, graph_scope: string}
    Literal: {lexical_value: string, datatype: IRI?, language: string?}
    Term: union[Resource, BlankNode, Literal]
    Predicate: {iri: IRI, label: string?, domain: IRI[]?, range: IRI[]?, inverse_of: IRI?, provenance: string?}
    Triple: {subject: union[Resource, BlankNode], predicate: Predicate, object: Term}
    NamedGraph: {id: IRI, triples: Triple[], provenance: string?, version: string?}
    Assertion: {triple: Triple, graph: IRI?, evidence: string[], confidence: number?, valid_from: datetime?, valid_to: datetime?}

  predicates:
    relates(subject: union[Resource, BlankNode], predicate: Predicate, object: Term)
    has_domain(predicate: Predicate, class: IRI)
    has_range(predicate: Predicate, class: IRI)
    subproperty_of(child: Predicate, parent: Predicate)
    inverse_of(a: Predicate, b: Predicate)
    equivalent_property(a: Predicate, b: Predicate)
    asserted_in(triple: Triple, graph: NamedGraph)

  constraints:
    - "An RDF predicate must be an IRI; literals and blank nodes cannot occupy the predicate position."
    - "A triple subject must be an IRI resource or blank node under RDF 1.1 conventions."
    - "Literal language tags and datatypes must be well formed; language-tagged strings require language-compatible datatypes."
    - "Property domain and range declarations entail types under RDFS; they are not closed-world validation constraints."
    - "Inverse, symmetric, transitive and functional behaviour requires explicit vocabulary axioms and a declared reasoning regime."
    - "Never equate two predicate IRIs merely because their labels match."
    - "Confidence annotations belong to assertions or evidence records, not to RDF truth semantics by default."

  inference:
    rule subproperty_propagation:
      if relates(s,p,o) and subproperty_of(p,q)
      then relates(s,q,o)
      condition: "RDFS entailment enabled"
    rule domain_typing:
      if relates(s,p,o) and has_domain(p,c)
      then type_of(s,c)
      condition: "RDFS entailment enabled"
    rule range_typing:
      if relates(s,p,o) and has_range(p,c) and is_resource_or_blank_node(o)
      then type_of(o,c)
      condition: "RDFS entailment enabled; avoid invalid literal-as-resource typing"
    rule inverse_relation:
      if relates(s,p,o) and inverse_of(p,q) and is_resource_or_blank_node(o)
      then relates(o,q,s)
      condition: "OWL-compatible inverse-property reasoning enabled"

  SEMANTIC_EQUIVALENCE:
    - "Distinguish equivalentProperty, subPropertyOf, inverseOf and informal lexical similarity."
    - "Normalise IRIs without conflating distinct namespace identities."
    - "Preserve ontology versions and the entailment regime used for equivalence claims."

  CONFIDENCE:
    range: [0,1]
    factors: [evidence_quality, source_reliability, mapping_quality, inference_assumptions, temporal_validity]
    policy: "Keep confidence separate from logical entailment and never manufacture certainty through chained assertions."

  VALIDATION:
    - "Check RDF term positions, IRI syntax, literal datatypes, graph scope and provenance."
    - "Use SHACL or an explicitly declared constraint language for application-level shape validation."
    - "Check property mappings for unintended identity merges and unsupported inverse or transitive assumptions."
    - "Report inconsistent claims without silently deleting conflicting evidence."

  FLOW_HASH:
    algorithm: "SHA-256"
    input: "canonicalised ontology version, named graph assertions, inference rules and provenance"
    note: "Canonicalise blank nodes using a defined RDF dataset canonicalisation procedure; compute hash at execution."

  UPDATE_MODEL:
    strategy: "versioned, provenance-preserving incremental update"
    steps: [ingest, parse, normalise, validate, reconcile, recompute_entailments, publish]
    rollback: true

  AI_CONSUMPTION:
    representation: "typed RDF knowledge graph with explicit predicate semantics"
    outputs: [predicate_catalogue, relation_triples, inferred_triples, provenance_records, uncertainty_annotations]
    safeguards: [preserve_predicate_IRIs, distinguish_inference_from_assertion, no_unsupported_equivalence, respect_graph_scope]
}

END FLOW ALGEBRA