Knowledge Graph Domain Ontology

FLOW ALGEBRA
TITLE: Knowledge Graph Domain Ontology
VERSION: 1.0
SOURCE_URL: https://kirethresearch.org/booksai-core/ontologies/domain-ontologies/information-domains/knowledge-graph-domain-ontology.html
STATUS: Domain-based draft; source page content not independently verified.
AUTHORS: Mark Higgins — Sky (Recursive AI)
PROJECT: Kireth Semantic Library

DOMAIN:
  id: knowledge_graph_domain_ontology
  parent: information_domains
  definition: A formal model for representing entities, concepts, properties and typed relationships as a graph of semantically interpretable assertions.

TYPES:
  Entity: identifiable thing represented in a knowledge graph
  Concept: class or abstract category
  Relationship: typed association between nodes
  Property: attribute of an entity or relationship
  Triple: subject-predicate-object assertion
  NamedGraph: identified collection of assertions
  Ontology: vocabulary of classes, predicates and constraints
  Identifier: persistent reference to a resource
  Literal: typed or language-tagged value
  ProvenanceRecord: source and derivation information
  Evidence: support for a claim
  GraphVersion: immutable or traceable graph state
  Inference: derived assertion with supporting premises
  Constraint: rule restricting valid graph structure
  ConfidenceScore: qualified uncertainty attached to an assertion

RELATIONSHIPS:
  instance_of(Entity, Concept)
  subclass_of(Concept, Concept)
  related_to(Entity, Entity)
  has_property(Entity, Property)
  asserts(NamedGraph, Triple)
  subject_of(Entity, Triple)
  predicate_of(Relationship, Triple)
  object_of(Entity|Literal, Triple)
  supported_by(Triple, Evidence)
  derived_from(Inference, Triple)
  sourced_from(Triple, ProvenanceRecord)
  supersedes(GraphVersion, GraphVersion)
  equivalent_to(Entity|Concept, Entity|Concept)
  contradicts(Triple, Triple)

CONSTRAINTS:
  - Each assertion must have an identifiable subject, predicate and object.
  - Distinguish asserted triples from inferred triples.
  - Preserve original source and time for every externally sourced assertion.
  - Never equate two entities solely because their labels match.
  - Apply ontology domain/range and cardinality constraints when declared.
  - Do not treat missing statements as false without an explicit closed-world assumption.
  - Conflicting assertions must remain inspectable until resolved.

INFERENCE_RULES:
  R1: IF instance_of(x,A) AND subclass_of(A,B) THEN instance_of(x,B).
  R2: IF subclass_of(A,B) AND subclass_of(B,C) THEN subclass_of(A,C).
  R3: IF equivalent_to(A,B) is validated THEN canonicalise identifiers while preserving aliases and provenance.
  R4: IF asserted_triple(t) AND valid_rule(r,t) THEN derived_triple(t2) WITH derivation(r,t).
  R5: IF contradicts(t1,t2) THEN flag_conflict(t1,t2); do not silently discard either.
  R6: IF source_is_revoked(s) THEN mark dependent assertions for revalidation.

CONFIDENCE:
  range: [0.0,1.0]
  semantics: calibrated estimate of support, not logical truth
  factors: [source_reliability, evidence_quality, recency, corroboration, inference_depth]
  propagation: explicit rule-dependent; never assume independent evidence
  unknown: null

SEMANTIC_EQUIVALENCE:
  levels: [exact_identity, contextual_equivalence, close_match, related_match, distinct]
  requirements: [type_compatibility, contextual_validation, provenance]
  caution: lexical similarity alone does not establish identity

PROVENANCE:
  fields: [source_uri, author_or_agent, observation_time, ingestion_time, method, licence, evidence_reference]
  preservation: retain original assertions and derivation chains

FLOW_HASH:
  algorithm: SHA-256
  input: canonical_serialisation(domain, version, ontology, assertions, rules, provenance)
  purpose: integrity and reproducible change detection
  note: hash value must be computed by an implementation; not fabricated

UPDATE_MODEL:
  operations: [add, revise, deprecate, retract, merge, split]
  procedure: validate -> version -> recalculate_inferences -> update_indexes -> audit
  preserve_history: true
  conflict_policy: retain evidence and request resolution

AI_CONSUMPTION:
  formats: [RDF, JSON-LD, Turtle, property_graph]
  capabilities: [entity_linking, semantic_search, retrieval_augmented_generation, explainable_reasoning]
  requirements: [typed_identifiers, evidence_tracing, confidence_visibility, version_awareness]
  safeguards: [avoid_unsupported_inferences, expose_uncertainty, respect_access_controls]

VALIDATION:
  checks: [identifier_uniqueness, schema_consistency, constraint_conformance, provenance_completeness, inference_traceability]
  outcome: pass | warning | fail

END FLOW ALGEBRA