Inference Graph Ontology

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

flow "InferenceGraphOntology" v1.0 {
  ontology:
    id: "inference-graph-ontology"
    parent: "ai-reasoning-relationships"
    purpose: "Represent evidence-backed propositions, inference rules and derivations as a typed, auditable reasoning graph."

  types:
    Proposition: {id: string, statement: string, context: string?, truth_status: enum[asserted, inferred, disputed, unknown], provenance: string[]}
    InferenceRule: {id: string, kind: enum[deductive, inductive, abductive, defeasible], premises: string[], conclusion: string, assumptions: string[], version: string}
    InferenceStep: {id: string, rule_id: string, premise_ids: string[], conclusion_id: string, timestamp: datetime?, confidence: number, evidence: string[]}
    Evidence: {id: string, source: string, reliability: number?, timestamp: datetime?, scope: string?}
    InferenceGraph: {id: string, proposition_ids: string[], step_ids: string[], context: string?, version: string}
    Relationship: {id: string, subject: string, predicate: string, object: string, confidence: number, provenance: string[]}

  predicates:
    premise_of(p: Proposition, step: InferenceStep)
    concludes(step: InferenceStep, p: Proposition)
    applies_rule(step: InferenceStep, rule: InferenceRule)
    supported_by(p: Proposition, evidence: Evidence)
    contradicts(a: Proposition, b: Proposition)
    depends_on(a: Proposition, b: Proposition)
    member_of(p: Proposition, graph: InferenceGraph)

  constraints:
    - "All referenced propositions, rules and evidence must resolve to typed records."
    - "Confidence and evidence reliability, where present, must lie within [0,1]."
    - "Every inferred conclusion must reference an explicit inference step and premises."
    - "Do not treat abductive or inductive conclusions as deductive entailments."
    - "Do not assume graph edges imply causal relationships."
    - "Detect cyclic dependencies; permit fixed-point reasoning only with an explicit termination or convergence policy."
    - "Preserve conflicting claims and provenance rather than silently overwriting them."

  inference:
    rule valid_modus_ponens:
      if asserted(p) and asserted(implies(p,q)) and same_context(p,q)
      then derive(q)
      evidence: "record both premises and the rule used"
    rule dependency_trace:
      if premise_of(p,s) and concludes(s,q)
      then depends_on(q,p)
      confidence: "do not manufacture certainty"
    rule contradiction_flag:
      if asserted(p) and asserted(not_p) and same_context(p,not_p)
      then flag_conflict(p,not_p)
      policy: "do not resolve without additional evidence"

  SEMANTIC_EQUIVALENCE:
    - "Normalise identifiers only against versioned, provenance-backed mappings."
    - "Distinguish logical equivalence, semantic similarity, identity and paraphrase."
    - "Preserve context, quantifiers, scope and negation when matching propositions."

  CONFIDENCE:
    range: [0,1]
    factors: [premise_quality, rule_validity, evidence_reliability, context_match, independent_support]
    policy: "Confidence aggregation requires a declared model; never increase confidence solely by chaining steps."

  VALIDATION:
    - "Check graph references, type compatibility, rule preconditions and provenance."
    - "Verify deductions against the selected logical semantics."
    - "Flag cycles, unsupported conclusions, contradictions and missing assumptions."
    - "Retain uncertainty and alternative explanations."

  FLOW_HASH:
    algorithm: "SHA-256"
    input: "canonicalised ontology version, graph nodes, edges, rules and provenance"
    note: "Compute hash at execution; no hash value is claimed here."

  UPDATE_MODEL:
    strategy: "versioned, provenance-preserving incremental graph update"
    steps: [ingest, normalise, validate, detect_changes, retract_stale_inferences, recompute_affected_paths, publish]
    rollback: true

  AI_CONSUMPTION:
    representation: "typed inference graph with explicit derivation and evidence links"
    outputs: [propositions, premise_links, inference_steps, conclusions, confidence_annotations, provenance_records, conflict_flags]
    safeguards: [no_unsupported_entailment, distinguish_reasoning_modes, preserve_context, trace_all_inferences]
}

END FLOW ALGEBRA