Decision Relationship Ontology

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

flow "DecisionRelationshipOntology" v1.0 {
  ontology:
    id: "decision-relationship-ontology"
    parent: "ai-reasoning-relationships"
    purpose: "Represent evidence-backed relationships among decision problems, alternatives, criteria, constraints, outcomes and justifications."

  types:
    DecisionProblem: {id: string, objective: string, context: string, decision_time: datetime?}
    Alternative: {id: string, problem_id: string, description: string}
    Criterion: {id: string, name: string, direction: enum[maximise, minimise, satisfy], weight: number?}
    Constraint: {id: string, expression: string, hard: boolean}
    Evidence: {id: string, source: string, observed_at: datetime?, reliability: number?, provenance: string}
    Outcome: {id: string, alternative_id: string, metric: string?, value: number?, unit: string?, probability: number?}
    Decision: {id: string, problem_id: string, selected_alternative_id: string, timestamp: datetime, rationale: string, evidence_ids: string[]}
    Relationship: {id: string, subject: string, predicate: string, object: string, context: string?, confidence: number, evidence: string[]}

  predicates:
    candidate_for(a: Alternative, p: DecisionProblem)
    evaluated_against(a: Alternative, c: Criterion)
    constrained_by(p: DecisionProblem, c: Constraint)
    supported_by(d: Decision, e: Evidence)
    selects(d: Decision, a: Alternative)
    predicts(a: Alternative, o: Outcome)
    preferred_to(a: Alternative, b: Alternative, context: string)
    justified_by(d: Decision, rationale: string)

  constraints:
    - "Relationship endpoints must resolve to compatible typed entities."
    - "Confidence, evidence reliability and outcome probabilities must lie within [0,1]."
    - "Every selected alternative must belong to the decision problem."
    - "Criteria, units, weights, constraints and evaluation context must be explicit."
    - "Hard constraints must be satisfied or a documented exception recorded."
    - "Preferences are context-dependent and are not automatically transitive."
    - "Predicted outcomes must not be represented as observed results."
    - "Correlation or temporal succession alone must not establish causation."
    - "A recommendation is not proof that an alternative is optimal."

  inference:
    rule eligible_alternative:
      if candidate_for(a,p) and satisfies_all_hard_constraints(a,p)
      then eligible_for(a,p)
      confidence: "retain constraint evidence and evaluation context"
    rule decision_support:
      if selects(d,a) and supported_by(d,e) and valid(e)
      then evidence_contributes_to(e,d)
      confidence: "dependent on evidence quality; no automatic certainty"
    rule dominance:
      if eligible_for(a,p) and eligible_for(b,p) and no_worse_on_all_criteria(a,b,p) and better_on_some_criterion(a,b,p)
      then pareto_dominates(a,b,p)
      confidence: "valid only for stated criteria, measurements and assumptions"

  SEMANTIC_EQUIVALENCE:
    - "Normalise decision, choice, selection and recommendation labels only where their formal roles match."
    - "Distinguish an alternative from a decision, a predicted outcome from an observed outcome, and evidence from a justification."
    - "Preserve context, objective, criteria and time when comparing decisions."

  CONFIDENCE:
    range: [0,1]
    factors: [evidence_quality, uncertainty, model_calibration, assumption_validity, context_stability, independent_replication]
    policy: "Do not increase confidence solely by chaining dependent inferences or aggregating repeated evidence."

  VALIDATION:
    - "Check entity types, relationship endpoints, referential integrity and provenance."
    - "Check units, probability ranges, missing criteria and conflicting constraints."
    - "Record unresolved trade-offs, uncertainty and alternative explanations."
    - "Require an auditable decision rationale and versioned input assumptions."

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

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

  AI_CONSUMPTION:
    representation: "typed decision knowledge graph with evidence-backed relationships"
    outputs: [decision_links, alternatives, criteria, constraints, evidence_traces, confidence_annotations, inferred_relationships]
    safeguards: [no_unsupported_optimality_claims, distinguish_predictions_from_observations, preserve_uncertainty, explain_selection]
}

END FLOW ALGEBRA