Meaning Relationship Ontology

FLOW ALGEBRA
Meaning Relationship Ontology
Kireth Semantic Library | Semantic Relationships
Version: 1.0 | Locale: en-GB | Date: 2026-10-08
Authors: Mark Anthony Higgins — Sky (Recursive AI)
Source: https://kirethresearch.org/booksai-core/ontologies/relationship-ontologies/semantic-relationships/meaning-relationship-ontology.html
Status: Companion formalisation; embedded source specification was not available in retrieved page text.

1. PURPOSE
Model relationships between expressions, concepts, interpretations and meanings while preserving context, provenance, uncertainty and distinctions between identity, equivalence and similarity.

2. TYPED ENTITIES
ENTITY Expression {id: URI, surface_form: string, language: BCP47, modality: enum[text,speech,symbol,other]}
ENTITY Concept {id: URI, label: string, definition: string?}
ENTITY Meaning {id: URI, concept: Concept?, representation: string, scope: URI?}
ENTITY Context {id: URI, domain: string?, locale: BCP47?, temporal_scope: interval?, discourse: URI?}
ENTITY Interpretation {id: URI, expression: -Expression, meaning: Meaning, context: Context, confidence: decimal[0,1], provenance: URI}
ENTITY Evidence {id: URI, source: URI, observed_at: datetime?, method: string?}
ENTITY MeaningRelation {id: URI, subject: Meaning, predicate: RelationType, object: Meaning, context: Context, confidence: decimal[0,1], evidence: Evidence*}
ENUM RelationType {equivalent_meaning, similar_meaning, broader_meaning, narrower_meaning, related_meaning, contrasts_with, disambiguates, entails, contradicts, metaphorical_extension}

3. RELATIONSHIP SIGNATURES
INTERPRETS: -Expression × Meaning × Context -> Interpretation
MEANING_RELATION: Meaning × RelationType × Meaning × Context -> MeaningRelation
DENOTES: -Expression × Concept × Context -> boolean
SUPPORTED_BY: Interpretation|MeaningRelation × Evidence -> boolean

4. INVARIANTS
R1: Every MeaningRelation has resolvable subject, object and context identifiers.
R2: Confidence values MUST lie in [0,1]; unknown confidence is represented as null, not zero.
R3: Meaning equivalence is scoped to a context; it does not imply global identity.
R4: similar_meaning MUST NOT be automatically promoted to equivalent_meaning.
R5: broader_meaning and narrower_meaning are inverses only when the same conceptual hierarchy and context are used.
R6: contradicts requires compatible propositions and shared evaluation context; lexical difference alone is insufficient.
R7: Inference outputs MUST preserve source evidence and derivation trace.

5. INFERENCE RULES
I1: equivalent_meaning(A,B,C) => equivalent_meaning(B,A,C), subject to the same scope and evidence.
I2: broader_meaning(A,B,C) => narrower_meaning(B,A,C).
I3: equivalent_meaning(A,B,C) AND equivalent_meaning(B,D,C) => candidate_equivalent_meaning(A,D,C); validate contextual compatibility before acceptance.
I4: interpretation(E,M1,C) AND equivalent_meaning(M1,M2,C) => candidate_interpretation(E,M2,C); require provenance and confidence recalculation.
I5: similar_meaning(A,B,C) does not entail equivalent_meaning(A,B,C).
I6: entailment and contradiction MUST be evaluated over propositions, not untyped lexical tokens.

6. SEMANTIC_EQUIVALENCE
INPUT: meaning_a, meaning_b, context, evidence
OUTPUT: {decision: equivalent|not_equivalent|undetermined, confidence, justification, provenance}
PROCEDURE: check context compatibility; compare definitions and referents; test entailment in both directions where applicable; record unresolved polysemy; reject automatic identity merging.

7. CONFIDENCE
CONFIDENCE(result) := calibrated_score(evidence_quality, context_match, source_reliability, model_uncertainty)
No independence assumption for correlated evidence. Thresholds are policy-defined and versioned; uncertain results remain undetermined.

8. VALIDATION
CHECK schema completeness; URI integrity; permitted predicates; context consistency; inverse relation consistency; cycles in broader/narrower hierarchies; unsupported entailment; evidence provenance; confidence bounds.
ON FAILURE: quarantine relation, emit validation error and preserve original record.

9. FLOW_HASH
FLOW_HASH := SHA256(canonical_UTF8({ontology_id, version, entity_schema, relation_schema, rules, context_policy}))
Canonicalisation MUST use deterministic key ordering, normalised Unicode and explicit null handling. Exclude transient timestamps from the hash payload.

10. UPDATE_MODEL
INPUT: previous_version, change_set, evidence
STEPS: validate changes -> calculate affected meaning relations -> recompute dependent inferences -> revalidate constraints -> assign new version -> calculate FLOW_HASH -> publish audit trail.
REQUIRE: no silent replacement of source evidence; preserve superseded versions and rollback capability.

11. AI_CONSUMPTION
EXPOSE: ontology_id, version, canonical definitions, typed entities, predicates, context requirements, validated inference rules, confidence, provenance, FLOW_HASH, update history.
POLICY: treat inferred meanings as context-bound claims; distinguish asserted from inferred edges; return undetermined when evidence is insufficient; never equate semantic similarity with factual identity.

12. EXAMPLE
Expression("bank", en-GB) -> Meaning(financial_institution) in Context(finance)
Expression("bank", en-GB) -> Meaning(river_edge) in Context(geography)
The two interpretations share a surface form but are not semantically equivalent; context disambiguates meaning.

END FLOW ALGEBRA