Reasoning Chain Ontology

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

flow "ReasoningChainOntology" v1.0 {
  ontology:
    id: "reasoning-chain-ontology"
    parent: "ai-reasoning-relationships"
    purpose: "Represent auditable chains of explicit premises, intermediate assertions, inference steps, and conclusions without treating generated explanations as verified proofs."

  types:
    Proposition: {id: string, statement: string, language: string?, context: string?, truth_status: enum[asserted, supported, refuted, unknown]}
    Evidence: {id: string, source: string, observed_at: datetime?, provenance: string, reliability: number?}
    InferenceRule: {id: string, name: string, logic: string, preconditions: string[], conclusion_pattern: string, soundness_status: enum[verified, assumed, unknown]}
    ReasoningStep: {id: string, premise_ids: string[], rule_id: string, conclusion_id: string, evidence_ids: string[], confidence: number, explanation: string?}
    ReasoningChain: {id: string, step_ids: string[], goal_id: string, context: string?, version: string, provenance: string}
    Assessment: {id: string, target_id: string, status: enum[valid, invalid, inconclusive], findings: string[], evaluator: string}

  predicates:
    uses_premise(step: ReasoningStep, premise: Proposition)
    applies_rule(step: ReasoningStep, rule: InferenceRule)
    derives(step: ReasoningStep, conclusion: Proposition)
    supported_by(proposition: Proposition, evidence: Evidence)
    precedes_step(first: ReasoningStep, second: ReasoningStep)
    belongs_to(step: ReasoningStep, chain: ReasoningChain)
    evaluates(assessment: Assessment, chain: ReasoningChain)

  constraints:
    - "Every reference must resolve to a typed entity."
    - "Confidence and evidence reliability, when present, must be within [0, 1]."
    - "Every reasoning step must declare its premises, rule and conclusion."
    - "A step can consume a derived proposition only after its producing step, unless a separately justified fixed-point semantics is declared."
    - "Acyclic chains must admit a topological ordering; cycles require explicit semantics and termination checks."
    - "Do not equate a plausible narrative with a valid deduction."
    - "Deductive validity, empirical support and calibrated confidence are distinct attributes."
    - "Contradictory premises must be flagged; classical explosion must not be silently applied."
    - "Provenance and contextual assumptions must remain attached to derived assertions."

  inference:
    rule modus_ponens:
      if supported(P) and supported(implies(P,Q)) and valid_rule("modus_ponens")
      then derive(Q)
      confidence: "retain premise evidence; do not manufacture certainty"
    rule step_dependency:
      if derives(step_a,P) and uses_premise(step_b,P)
      then precedes_step(step_a,step_b)
      confidence: "structural, conditional on resolved identities"
    rule dependency_transitivity:
      if precedes_step(a,b) and precedes_step(b,c) and acyclic_chain(a,b,c)
      then precedes_step(a,c)
      confidence: "structural, not proof of conclusion truth"

  SEMANTIC_EQUIVALENCE:
    - "Normalise proposition identifiers only within an explicitly declared logical and contextual scope."
    - "Distinguish paraphrase, logical equivalence, entailment and empirical agreement."
    - "Do not merge steps solely because their natural-language explanations resemble one another."

  CONFIDENCE:
    range: [0,1]
    factors: [premise_support, evidence_reliability, rule_validity, uncertainty, context_fit, independent_verification]
    policy: "Propagate uncertainty conservatively; never raise confidence merely because a chain is longer."

  VALIDATION:
    - "Resolve all entity, premise, evidence and rule references."
    - "Check each rule application against declared preconditions."
    - "Detect missing premises, unsupported leaps, circular justification and inconsistent contexts."
    - "Separate syntactic well-formedness, deductive validity and factual soundness."
    - "Preserve unresolved alternatives and explicit uncertainty."

  FLOW_HASH:
    algorithm: "SHA-256"
    input: "canonicalised ontology version, chain structure, rules, assertions and provenance"
    note: "Compute at execution; no precomputed hash is claimed."

  UPDATE_MODEL:
    strategy: "versioned, provenance-preserving incremental revision"
    steps: [ingest, normalise, validate, identify_affected_steps, recompute_dependencies, reassess_conclusions, publish]
    rollback: true

  AI_CONSUMPTION:
    representation: "typed reasoning dependency graph with evidence-backed inference edges"
    outputs: [premise_links, step_dependencies, conclusion_records, validity_assessments, confidence_annotations, provenance_records]
    safeguards: [no_fabricated_proofs, no_unsupported_conclusions, distinguish_evidence_from_inference, preserve_context, disclose_uncertainty]
}

END FLOW ALGEBRA