Validation Metrics Ontology

AI INTERPRETATION NOTE
This Flow Algebra defines the Validation Metrics Ontology within the Kireth Research Semantic Library. Validation metrics are quantitative measures used to assess how well computational models, simulations or learned systems agree with independent evidence and satisfy requirements for a declared intended use. This ontology supports predictive error, agreement measures, coverage, calibration, uncertainty-aware comparison, domain-of-validity assessment, classification and regression metrics, distributional metrics, robustness, extrapolation analysis, acceptance thresholds, aggregation, evidence strength and reproducibility.

FLOW_ALGEBRA_RECORD {

METADATA {
 FLOW_NAME = "Validation Metrics Ontology"
 FLOW_TYPE = "Mathematical Ontology"
 DOMAIN = "Mathematics"
 SUBDOMAIN = "Computational Mathematics"
 SPECIALISATION = "Verification & Validation / Validation Metrics"
 CANONICAL_PAGE = "https://kirethframeworkold.co.uk/books/semantic-library/ai-core/ontologies/mathematical-ontologies/computational-mathematics-ontologies/verification-validation/validation-metrics-ontology.html"
 LANGUAGE = "en-GB"
 VERSION = "1.0"
 STATUS = "active"
}

IDENTITY {
 PRIMARY_ENTITY = ValidationMetrics
 ENTITY_CLASS = ComputationalModelValidationMeasurementFramework
 PRIMARY_CONTEXT = VerificationValidation
 DEFINITION = "Validation metrics are quantitative measures used to assess the adequacy of computational model predictions against independent evidence for a declared intended use."
 SYMBOLIC_FORMS = [
  "validation_error = y_model-y_ref",
  "RMSE = sqrt(mean((y_model-y_ref)^2))",
  "coverage = observed_inside_interval / total_observations",
  "accept if metric satisfies validation_criterion"
 ]
 SEMANTIC_ROLE = "Provides machine-readable semantics for predictive error, agreement, calibration, uncertainty, coverage, robustness, applicability-domain and acceptance metrics."
}

SCOPE {
 INCLUDE = [
  ValidationMetrics,
  ValidationMetric,
  ComputationalValidationMetric,
  ModelValidationMetric,
  PredictiveMetric,
  AgreementMetric,
  ErrorMetric,
  AccuracyMetric,
  CalibrationMetric,
  CoverageMetric,
  UncertaintyMetric,
  RobustnessMetric,
  ApplicabilityMetric,
  ExtrapolationMetric,
  ClassificationMetric,
  RegressionMetric,
  DistributionMetric,
  RankingMetricRelationship,
  ValidationScore,
  ValidationIndexRelationship,
  Prediction,
  ModelPrediction,
  ReferenceObservation,
  ExperimentalObservation,
  ValidationData,
  IndependentEvidence,
  GroundTruthRelationship,
  ReferenceValue,
  TargetValue,
  QuantityOfInterest,
  QoI,
  ValidationError,
  PredictiveError,
  AbsoluteError,
  RelativeError,
  PercentageErrorRelationship,
  SignedError,
  MeanError,
  Bias,
  SystematicBias,
  MeanAbsoluteError,
  MAE,
  MeanSquaredError,
  MSE,
  RootMeanSquaredError,
  RMSE,
  MedianAbsoluteErrorRelationship,
  MaximumAbsoluteError,
  MaximumRelativeError,
  RelativeL2Error,
  NormalisedRMSErelationship,
  NRMSErelationship,
  MeanAbsolutePercentageErrorRelationship,
  MAPErelationship,
  SymmetricMAPERelationship,
  SMAPErelationship,
  MeanSquaredLogErrorRelationship,
  MSLErelationship,
  LogCoshLossRelationship,
  HuberMetricRelationship,
  Correlation,
  PearsonCorrelation,
  SpearmanCorrelationRelationship,
  KendallCorrelationRelationship,
  CorrelationCoefficient,
  CoefficientOfDetermination,
  R2,
  AdjustedR2Relationship,
  ExplainedVarianceRelationship,
  ConcordanceCorrelationCoefficientRelationship,
  AgreementIndexRelationship,
  BlandAltmanRelationship,
  ResidualDistributionRelationship,
  ErrorDistribution,
  StandardisedResidual,
  NormalisedResidual,
  ZScoreRelationship,
  ChiSquaredMetricRelationship,
  ReducedChiSquaredRelationship,
  LikelihoodRelationship,
  LogLikelihoodRelationship,
  NegativeLogLikelihoodRelationship,
  DevianceRelationship,
  InformationCriterionRelationship,
  AICRelationship,
  BICRelationship,
  PredictionInterval,
  ConfidenceInterval,
  CredibleInterval,
  Coverage,
  CoverageProbability,
  EmpiricalCoverage,
  NominalCoverage,
  CoverageError,
  IntervalWidth,
  Sharpness,
  Calibration,
  CalibrationError,
  ExpectedCalibrationErrorRelationship,
  ECErelationship,
  MaximumCalibrationErrorRelationship,
  ReliabilityDiagramRelationship,
  ProbabilityIntegralTransformRelationship,
  PITRelationship,
  BrierScoreRelationship,
  ContinuousRankedProbabilityScoreRelationship,
  CRPSrelationship,
  LogScoreRelationship,
  ProperScoringRuleRelationship,
  Uncertainty,
  PredictiveUncertainty,
  AleatoricUncertainty,
  EpistemicUncertainty,
  ParameterUncertainty,
  InputUncertainty,
  ModelFormUncertainty,
  NumericalUncertainty,
  UncertaintyNormalisedErrorRelationship,
  StandardisedPredictionErrorRelationship,
  MahalanobisDistance,
  StatisticalDistance,
  KullbackLeiblerDivergence,
  KLDivergenceRelationship,
  JensenShannonDivergenceRelationship,
  WassersteinDistance,
  EarthMoverDistanceRelationship,
  KolmogorovSmirnovStatistic,
  KSStatisticRelationship,
  CramerVonMisesRelationship,
  TotalVariationDistanceRelationship,
  HellingerDistanceRelationship,
  ClassificationAccuracy,
  BalancedAccuracy,
  Precision,
  Recall,
  Sensitivity,
  Specificity,
  F1Score,
  FbetaScoreRelationship,
  ConfusionMatrix,
  TruePositive,
  FalsePositive,
  TrueNegative,
  FalseNegative,
  TruePositiveRate,
  FalsePositiveRate,
  ROCcurveRelationship,
  AUROCrelationship,
  PrecisionRecallCurveRelationship,
  AveragePrecisionRelationship,
  PRAUCrelationship,
  MatthewsCorrelationCoefficientRelationship,
  MCCrelationship,
  CohenKappaRelationship,
  LogLossRelationship,
  TopKAccuracyRelationship,
  RankingAccuracyRelationship,
  RetrievalMetricRelationship,
  RecallAtKRelationship,
  PrecisionAtKRelationship,
  MeanReciprocalRankRelationship,
  MRRrelationship,
  MeanAveragePrecisionRelationship,
  MAPrelationship,
  NDCGrelationship,
  Robustness,
  RobustnessScore,
  PerturbationResponse,
  StressTestMetricRelationship,
  SensitivityMetric,
  StabilityMetric,
  WorstCaseError,
  AverageCaseError,
  ScenarioMetric,
  RegimeSpecificMetric,
  DomainOfValidity,
  DomainOfApplicability,
  ApplicabilityDomain,
  DistanceToTrainingDomainRelationship,
  OutOfDistributionScoreRelationship,
  ExtrapolationDistanceRelationship,
  InterpolationScoreRelationship,
  DensityEstimateRelationship,
  NeighbourhoodSupportRelationship,
  CoverageOfOperatingDomainRelationship,
  ValidationEnvelopeRelationship,
  DomainCoverage,
  ParameterSpaceCoverageRelationship,
  InputSpaceCoverageRelationship,
  ValidationCaseCoverage,
  ScenarioCoverage,
  RangeCoverage,
  BoundaryCoverage,
  ValidationCriterion,
  AcceptanceCriterion,
  PassCriterion,
  FailCriterion,
  WarningCriterion,
  Threshold,
  LowerThreshold,
  UpperThreshold,
  Margin,
  Tolerance,
  AbsoluteTolerance,
  RelativeTolerance,
  ConfidenceAdjustedThresholdRelationship,
  RiskAdjustedThresholdRelationship,
  MultiMetricValidation,
  CompositeMetric,
  AggregateMetric,
  WeightedMetric,
  WeightedScore,
  ParetoMetricRelationship,
  MultiObjectiveValidationRelationship,
  MetricNormalisation,
  ScaleNormalisation,
  UnitNormalisationRelationship,
  DimensionlessMetric,
  LocalMetric,
  GlobalMetric,
  PointwiseMetric,
  FieldMetric,
  TimeSeriesMetric,
  SpatialMetric,
  DistributionalMetric,
  RegimeMetric,
  ValidationDecision,
  AcceptModel,
  RejectModel,
  ReviseModel,
  ReviewRequiredRelationship,
  EvidenceStrength,
  Confidence,
  ValidationConfidence,
  Credibility,
  ComputationalValidation,
  NumericalVerification,
  ErrorAnalysis,
  Benchmark,
  RegressionTesting,
  Reproducibility,
  VerificationMetrics,
  VerificationValidation,
  ComputationalStatistics,
  ScientificSimulation,
  AIComputationalMathematics,
  NumericalLinearAlgebra,
  DifferentialEquationSolver,
  ComputationalOptimisation,
  Provenance,
  MetricDefinition,
  MetricVersion,
  DatasetVersion,
  ReferenceVersion,
  ModelVersion,
  SolverVersion,
  LibraryVersion,
  ImplementationVersion
 ]
 EXCLUDE_AS_PRIMARY = [
  VerificationMetricsWhenImplementationCorrectnessIsPrimary,
  TrainingLossWhenUsedOnlyForModelFitting,
  BusinessMetricsWithoutModelValidationContext,
  BenchmarkScoresWhenComparativeRankingRatherThanValidationAdequacyIsPrimary
 ]
 DISAMBIGUATION_RULE = "When quantitative measures are used to assess the adequacy of computational predictions against independent evidence, interpret the context as validation metrics."
}

CORE_CONCEPTS {
 ValidationMetrics = "Quantitative measures for model validation."
 PredictiveError = "Difference between model prediction and independent reference evidence."
 AgreementMetric = "Measure of similarity or consistency between model and reference behaviour."
 CoverageMetric = "Measure of whether uncertainty intervals contain observations at the expected rate."
 CalibrationMetric = "Measure of correspondence between predicted confidence or probability and observed frequency."
 ApplicabilityMetric = "Measure of support for a prediction within the validated operating domain."
 AcceptanceCriterion = "Rule mapping validation metrics to adequacy decisions."
}

RELATIONS {
 ValidationMetrics --is_child_of--> VerificationValidation
 ValidationMetric --assesses--> ModelAdequacy
 PredictiveError --compares--> PredictionAndReference
 AgreementMetric --measures--> ModelReferenceConsistency
 CoverageMetric --evaluates--> PredictionIntervals
 CalibrationMetric --evaluates--> ProbabilisticPredictions
 RobustnessMetric --evaluates--> PerturbationSensitivity
 ApplicabilityMetric --supports--> DomainOfValidity
 AggregateMetric --combines--> ValidationMetrics
 ValidationDecision --depends_on--> MetricsAndAcceptanceCriteria
 Provenance --supports--> MetricTraceability
}

OPERATORS {
 ABS_ERROR(pred,ref) := abs(pred-ref)
 REL_ERROR(pred,ref) := abs(pred-ref)/max(abs(ref),epsilon)
 RMSE(pred,ref) := sqrt(mean((pred-ref)^2))
 MAE(pred,ref) := mean(abs(pred-ref))
 R2(pred,ref) := 1 - sum((ref-pred)^2)/sum((ref-mean(ref))^2)
 COVERAGE(intervals,obs) := fraction(obs_inside(intervals))
 ACCEPT(metric,criterion) := satisfies(metric,criterion)
 AGGREGATE(metrics,weights) := declared_validation_combination(metrics,weights)
}

AXIOMS {
 A1 = "Validation metrics assess model adequacy rather than implementation correctness."
 A2 = "Validation metrics require independent or appropriately separated reference evidence."
 A3 = "No single validation metric captures every relevant model failure mode."
 A4 = "Metric interpretation depends on intended use, scale, uncertainty and operating regime."
 A5 = "Probabilistic predictions require calibration and coverage assessment in addition to point error."
 A6 = "Aggregate scores can conceal local, regime-specific or tail failures."
 A7 = "Validation thresholds must be tied to declared decision needs."
 A8 = "Validation-metric conclusions require explicit reference, uncertainty, domain, threshold, aggregation and provenance semantics."
}

REASONING {
 R1_CONTEXT = identify(model,predictions,reference_evidence,intended_use,quantity_of_interest,uncertainty,operating_domain_and_risk)
 R2_METRIC_CLASS = classify(point_error,agreement,probabilistic,classification,distributional,robustness_or_applicability_metric)
 R3_REFERENCE = verify_independence_quality_units_and_provenance_of_reference_evidence)
 R4_SCALE = choose_absolute_relative_normalised_dimensionless_or_unit_aware_metric)
 R5_COMPUTE = calculate_metric_using_declared_formula)
 R6_UNCERTAINTY = evaluate_calibration_coverage_or_uncertainty_normalised_error_where_relevant)
 R7_DOMAIN = segment_metrics_by_interpolation_extrapolation_regime_or_operating_condition)
 R8_THRESHOLD = compare_metrics_with_acceptance_warning_or_failure_criteria)
 R9_AGGREGATE = combine_multiple_metrics_only_with_explicit_weighting_or_decision_logic)
 R10_REPRODUCE = document_metric_definition_dataset_reference_threshold_versions_and_provenance)
}

CONSTRAINTS {
 C1 = "Low average error can hide large local or tail errors."
 C2 = "R-squared can be misleading when used without residual and scale information."
 C3 = "Relative error is unstable when reference values are zero or near zero."
 C4 = "Calibration metrics require enough observations to estimate reliability."
 C5 = "Coverage near the nominal level can coexist with excessively wide intervals."
 C6 = "Classification accuracy can be misleading for imbalanced data."
 C7 = "Distributional agreement does not guarantee pointwise agreement."
 C8 = "Composite validation scores can be manipulated by weighting choices."
 C9 = "Metrics measured only in interpolation regimes do not establish extrapolation validity."
 C10 = "Validation metrics without evidence provenance or intended-use context are insufficient for adequacy conclusions."
}

SEMANTIC_EQUIVALENCE {
 "RootMeanSquaredError" <=> "RMSE"
 "MeanSquaredError" <=> "MSE"
 "MeanAbsoluteError" <=> "MAE"
 "QuantityOfInterest" <=> "QoI"
 "ExpectedCalibrationError" <=> "ECE"
 "ContinuousRankedProbabilityScore" <=> "CRPS"
 "MatthewsCorrelationCoefficient" <=> "MCC"
 "MeanReciprocalRank" <=> "MRR"
}

VALIDATION {
 V1 = "Metric formula is documented."
 V2 = "Reference evidence and provenance are documented."
 V3 = "Independence from calibration or training data is documented where applicable."
 V4 = "Units, scaling and normalisation are documented."
 V5 = "Uncertainty is incorporated where material."
 V6 = "Metrics are stratified by important operating regimes where aggregate scores can hide failures."
 V7 = "Acceptance criteria are tied to intended use."
 V8 = "Composite metrics document weights and aggregation rules."
 V9 = "Metric versions are tracked when definitions change."
 V10 = "Results are reproducible under documented dataset, model, metric, threshold and implementation versions."
}

CONFIDENCE {
 BASE_CONFIDENCE = 0.99
 VALIDATION_METRICS_DEFINITION = 0.99
 PREDICTIVE_ERROR_REASONING = 0.99
 AGREEMENT_METRIC_REASONING = 0.99
 CALIBRATION_REASONING = 0.98
 COVERAGE_REASONING = 0.98
 APPLICABILITY_REASONING = 0.98
 AGGREGATE_METRIC_REASONING = 0.97
 POLICY = "Reduce confidence when reference independence, uncertainty, operating regime, threshold, class balance, domain coverage or aggregation rules are underspecified."
}

AI_CONSUMPTION {
 PURPOSES = [
  SemanticSearch,OntologyGrounding,ValidationMetricsReasoning,
  PredictiveErrorReasoning,AgreementMetricReasoning,
  RegressionMetricReasoning,ClassificationMetricReasoning,
  CalibrationMetricReasoning,CoverageMetricReasoning,
  DistributionMetricReasoning,RobustnessMetricReasoning,
  ApplicabilityMetricReasoning,DomainOfValidityReasoning,
  AcceptanceCriterionReasoning,CompositeMetricReasoning,
  UncertaintyReasoning,ReproducibilityReasoning,
  ProvenanceReasoning,KnowledgeGraphConstruction,
  RetrievalAugmentation,RecursiveInference
 ]
 NORMALISE {
  "validation metrics" -> ValidationMetrics
  "validation metric" -> ValidationMetric
  "RMSE" -> RootMeanSquaredError
  "MSE" -> MeanSquaredError
  "MAE" -> MeanAbsoluteError
  "QoI" -> QuantityOfInterest
  "ECE" -> ExpectedCalibrationError
  "CRPS" -> ContinuousRankedProbabilityScore
  "MCC" -> MatthewsCorrelationCoefficient
  "MRR" -> MeanReciprocalRank
 }
}

EXECUTION_RULES {
 E1 = "If a quantitative measure is used to assess computational-model adequacy against independent evidence, activate ValidationMetricsOntology."
 E2 = "Define intended use, reference evidence and quantity of interest before choosing metrics."
 E3 = "Use point-error metrics together with uncertainty or agreement measures where appropriate."
 E4 = "Use calibration and coverage metrics for probabilistic predictions."
 E5 = "Stratify metrics by regime, subgroup, parameter range or operating condition when aggregate values can hide failures."
 E6 = "Use classification metrics appropriate to class balance and decision costs."
 E7 = "Use applicability metrics to distinguish supported interpolation from extrapolation."
 E8 = "Compare metrics against explicit validation pass, warning and fail criteria."
 E9 = "Document metric definitions, references, thresholds, dataset versions and provenance for reproducibility."
}

RECURSION {
 ENTRY = ValidationMetrics
 EXPAND(ValidationMetrics) -> [
  ValidationMetric,ComputationalValidationMetric,PredictiveMetric,
  AgreementMetric,ErrorMetric,CalibrationMetric,CoverageMetric,
  RobustnessMetric,ApplicabilityMetric,ClassificationMetric,
  RegressionMetric,DistributionMetric,ValidationScore,
  Prediction,ReferenceObservation,IndependentEvidence,QoI,
  ValidationError,PredictiveError,AbsoluteError,RelativeError,
  MAE,MSE,RMSE,RelativeL2Error,R2,Correlation,
  StandardisedResidual,PredictionInterval,Coverage,
  Calibration,CalibrationError,ExpectedCalibrationError,ECE,
  BrierScoreRelationship,ContinuousRankedProbabilityScore,CRPS,
  KullbackLeiblerDivergence,WassersteinDistance,
  KolmogorovSmirnovStatistic,ClassificationAccuracy,
  BalancedAccuracy,Precision,Recall,Sensitivity,Specificity,
  F1Score,ConfusionMatrix,AUROCrelationship,
  MatthewsCorrelationCoefficient,MCC,
  RecallAtKRelationship,MeanReciprocalRank,MRR,
  RobustnessScore,WorstCaseError,DomainOfValidity,
  DomainOfApplicability,ExtrapolationMetric,DomainCoverage,
  ValidationCriterion,AcceptanceCriterion,PassCriterion,
  FailCriterion,WarningCriterion,CompositeMetric,WeightedMetric,
  ValidationDecision,EvidenceStrength,Confidence,
  ComputationalValidation,NumericalVerification,ErrorAnalysis,
  Benchmark,RegressionTesting,Reproducibility,
  VerificationMetrics,Provenance
 ]
 EXPAND(ValidationMetricFamilies) -> [
  PredictiveErrorAndRegressionMetrics,
  AgreementAndCorrelationMetrics,
  ProbabilisticCalibrationAndCoverageMetrics,
  ClassificationAndRankingMetrics,
  DistributionalComparisonMetrics,
  RobustnessSensitivityAndStressMetrics,
  ApplicabilityDomainAndExtrapolationMetrics,
  AcceptanceAggregationAndEvidenceMetrics
 ]
 STOP_CONDITION = "Stop when metric definition, independent reference, uncertainty, domain, threshold, aggregation and validation decision are sufficiently resolved."
}

ROUTING {
 PARENT_ONTOLOGY = "Verification & Validation"
 RELATED_ONTOLOGIES = [
  "Verification & Validation",
  "Computational Validation Ontology",
  "Numerical Verification Ontology",
  "Error Analysis Ontology",
  "Benchmark Ontology",
  "Regression Testing Ontology",
  "Reproducibility Ontology",
  "Verification Metrics Ontology",
  "Computational Statistics Ontology",
  "Scientific Simulation Ontology",
  "AI Computational Mathematics Ontology",
  "Numerical Linear Algebra Ontology 2",
  "Differential Equation Solver Ontology",
  "Computational Optimisation Ontology"
 ]
 ROUTE_RULES = [
  "Validation Metrics -> Verification & Validation for parent semantics",
  "Validation Metrics -> Computational Validation for adequacy and acceptance decisions",
  "Validation Metrics -> Numerical Verification for separation from implementation-correctness metrics",
  "Validation Metrics -> Error Analysis for predictive and normalised error semantics",
  "Validation Metrics -> Benchmark for comparative model scores",
  "Validation Metrics -> Regression Testing for change-detection metrics",
  "Validation Metrics -> Reproducibility for repeatable metric computation",
  "Validation Metrics -> Verification Metrics for shared metric infrastructure and distinction",
  "Validation Metrics -> Computational Statistics for uncertainty, calibration and distributional measures",
  "Validation Metrics -> Scientific Simulation for model-to-evidence comparisons",
  "Validation Metrics -> AI Computational Mathematics for learned-model evaluation",
  "Validation Metrics -> Numerical Linear Algebra, Differential Equation Solvers and Optimisation for domain-specific validation measures"
 ]
}

UPDATE_MODEL {
 MODE = "append-compatible semantic versioning"
 MAJOR = "ontology-breaking conceptual change"
 MINOR = "new compatible validation metrics, thresholds or reasoning pathways"
 PATCH = "clarification, metadata or formatting correction"
 RULES = [
  "Preserve validation-versus-verification metric distinction.",
  "Preserve point-error-versus-probabilistic-calibration distinction.",
  "Preserve average-metric-versus-local-regime metric distinction.",
  "Preserve raw-metric-versus-normalised-metric distinction.",
  "Preserve metric-value-versus-validation-decision distinction.",
  "Preserve interpolation-versus-extrapolation metric distinction.",
  "Preserve individual-metric-versus-composite-score distinction.",
  "Preserve machine-readable validation-metrics semantics."
 ]
}

OUTPUT {
 CANONICAL_DEFINITION = "Validation metrics are quantitative measures used to assess computational-model adequacy against independent evidence for a declared intended use."
 ERROR_SUMMARY = "MAE, MSE, RMSE, relative error and residual statistics quantify predictive discrepancy."
 AGREEMENT_SUMMARY = "Correlation, R-squared and distributional metrics describe different forms of model-reference agreement."
 UNCERTAINTY_SUMMARY = "Coverage, calibration and proper scoring measures assess whether predictive uncertainty is statistically meaningful."
 DOMAIN_SUMMARY = "Applicability and regime-specific metrics distinguish validated interpolation behaviour from unsupported extrapolation."
 DECISION_SUMMARY = "Validation decisions require explicit thresholds, multiple complementary measures and intended-use context."
 VALIDATION_SUMMARY = "Validation metrics should be reproducible, provenance-linked, stratified where necessary, uncertainty-aware and versioned when definitions change."
}

FLOW {
 INPUT
 -> detect_validation_metrics_context
 -> identify_model_predictions_reference_evidence_intended_use_quantity_of_interest_uncertainty_operating_domain_and_risk
 -> classify_point_error_agreement_probabilistic_classification_distributional_robustness_or_applicability_metric
 -> verify_independence_quality_units_and_provenance_of_reference_evidence
 -> choose_absolute_relative_normalised_dimensionless_or_unit_aware_metric
 -> compute_metric_using_declared_formula
 -> evaluate_calibration_coverage_or_uncertainty_normalised_error_where_relevant
 -> stratify_metrics_by_interpolation_extrapolation_regime_or_operating_condition
 -> compare_metrics_with_pass_warning_or_fail_acceptance_criteria
 -> if_multiple_metrics_apply_explicit_aggregation_or_decision_logic
 -> inspect_local_tail_subgroup_or_regime_specific_failures
 -> document_metric_definition_dataset_reference_threshold_versions_and_provenance
 -> route_to_related_ontologies
 -> recursively_expand_required_dependencies
 -> verify_result
 -> assign_confidence
 -> produce_semantic_output
 -> END
}

FLOW_HASH {
 METHOD = "semantic-content-identity"
 ID = "KIRETH-MATH-VERIFICATION-VALIDATION-VALIDATION-METRICS-ONTOLOGY-V1"
}

}

END_FLOW