Scientific Data Quality Ontology

# Scientific Data Quality Ontology — Flow Algebra

URL: https://kirethresearch.org/booksai-core/ontologies/mathematical-ontologies/measurement-units-ontologies/ai-measurement-semantics/scientific-data-quality-ontology.html

FLOW ScientificDataQualityOntology {

  DOMAIN:
    Mathematical Ontologies
    -> Measurement & Units Ontologies
    -> AI Measurement Semantics
    -> Scientific Data Quality Ontology

  PURPOSE:
    Represent scientific data quality as a multidimensional, evidence-based and machine-interpretable assessment of whether scientific measurement data are sufficiently accurate in representation, complete, consistent, timely, traceable, reproducible, valid, uncertainty-aware, provenance-linked and fit for a declared scientific use. The ontology models quality dimensions, measurement semantics, metadata completeness, uncertainty, calibration, traceability, provenance, integrity, missingness, censoring, outliers, bias, precision, sampling, temporal and spatial coverage, transformation lineage, statistical quality, validation, standards conformance, FAIR-related interoperability, AI-generated or inferred data, quality flags, confidence, quality rules, recursive re-evaluation and machine-consumable reasoning. It explicitly prevents data quality from being reduced to a single score or confused with measurement accuracy, precision, uncertainty, integrity, completeness or provenance.

  ENTITY ScientificDataQuality {
    TYPE: ScientificInformationQualityFramework

    PROPERTIES:
      has_quality_dimensions
      has_quality_requirements
      has_quality_evidence
      has_quality_rules
      has_measurement_context
      has_uncertainty_context
      has_provenance
      has_validation
      has_quality_flags
      has_intended_use
      supports_recursive_assessment
  }

  ENTITY ScientificDataset {
    TYPE: ScientificDataCollection

    COMPONENTS:
      dataset_identifier
      observations
      variables
      metadata
      sampling_context
      provenance
      version
      quality_status
  }

  ENTITY DataQualityDimension {
    TYPE: QualityCharacteristic

    POSSIBLE:
      semantic_correctness
      completeness
      consistency
      validity
      accuracy_related_quality
      precision_related_quality
      uncertainty_quality
      traceability
      provenance
      timeliness
      spatial_coverage
      temporal_coverage
      representativeness
      integrity
      reproducibility
      interoperability
  }

  ENTITY QualityRequirement {
    TYPE: DeclaredFitnessCriterion

    COMPONENTS:
      requirement_identifier
      quality_dimension
      target
      threshold_orcondition
      intended_use
      source
  }

  ENTITY QualityEvidence {
    TYPE: SupportingScientificEvidence

    POSSIBLE:
      calibration_record
      uncertainty_budget
      validation_result
      reference_measurement
      provenance_record
      statistical_test
      metadata_record
      audit_record
      comparison_result
  }

  ENTITY QualityAssessment {
    TYPE: ScopedEvaluation

    COMPONENTS:
      assessment_identifier
      target
      intended_use
      rule_set
      evidence
      results
      flags
      confidence
      provenance
  }

  QUALITY_DIMENSION_MODEL {
    dimensions:
      semantic
      metrological
      statistical
      contextual
      provenance
      integrity
      temporal
      spatial
      computational
      interoperability

    RULE:
      quality_is_multidimensional_and SHOULD not_be collapsed_to_single_unqualified_metric
  }

  FITNESS_FOR_USE_MODEL {
    quality:
      sufficient_for_declared_scientific_purpose

    RULE:
      high_quality_for_one_use MAY be inadequate_foranother
  }

  INTENDED_USE_MODEL {
    fields:
      scientific_question
      analysis
      required_uncertainty
      required_resolution
      population
      spatial_scale
      temporal_scale
      decision_context

    RULE:
      quality_assessment_without_intended_use_is_incomplete
  }

  SEMANTIC_QUALITY_MODEL {
    checks:
      variable_identity
      quantity_kind
      measurand
      unit
      dimension
      definitions
      coding

    RULE:
      syntactically_valid_data MAY be semantically_invalid
  }

  QUANTITY_KIND_MODEL {
    variable:
      semantic_quantity_type

    RULE:
      quantity_kind != dimension
  }

  UNIT_QUALITY_MODEL {
    checks:
      unit_identifier
      symbol
      prefix
      system
      conversion

    RULE:
      missing_orambiguous_unit_reduces_scientific_reusability
  }

  DIMENSIONAL_QUALITY_MODEL {
    checks:
      unit_dimension
      quantity_dimension
      equation_consistency

    RULE:
      dimensional_consistency_is_necessary_butnot_sufficient
  }

  METADATA_COMPLETENESS_MODEL {
    expected_fields:
      identifiers
      variables
      units
      methods
      instruments
      calibration
      uncertainty
      time
      location
      provenance

    completeness:
      present_required_fields/required_fields
      under_declared_schema

    RULE:
      completeness_score_depends_ondeclared_requirements
  }

  REQUIRED_METADATA_MODEL {
    requirement:
      field_required_forintended_use

    RULE:
      optional_in_schema MAY still_be_required_forparticular_analysis
  }

  MISSING_METADATA_MODEL {
    field:
      absent_orunknown

    RULE:
      missing_metadata MUST not_be silently_inferred_asfact
  }

  DATA_COMPLETENESS_MODEL {
    expected_observations:
      N_expected

    valid_observations:
      N_valid

    completeness:
      N_valid/N_expected
      WHEN N_expected > 0

    RULE:
      expected_observation_definition MUST be explicit
  }

  MISSING_DATA_MODEL {
    statuses:
      missing
      not_observed
      lost
      invalid
      censored
      below_detection
      above_range
      not_applicable

    RULE:
      missing != zero
  }

  MISSINGNESS_MECHANISM_MODEL {
    possible:
      MCAR
      MAR
      MNAR
      unknown

    RULE:
      missingness_mechanism affects_statistical_validity
  }

  IMPUTATION_MODEL {
    possible:
      deterministic
      statistical
      multiple
      model_based

    RULE:
      imputed_values MUST be flagged_andlineage_preserved
  }

  CENSORING_MODEL {
    possible:
      left
      right
      interval

    RULE:
      censored_data MUST retain_threshold_andstatus
  }

  DETECTION_LIMIT_MODEL {
    threshold:
      method_specific

    RULE:
      below_detection != zero
  }

  QUANTIFICATION_LIMIT_MODEL {
    threshold:
      level_forquantitative_reporting_underdeclared_performance

    RULE:
      detection_andquantification_limits MUST be distinguished
  }

  VALIDITY_MODEL {
    data:
      conforms_to semantic_physical_method_andrange_constraints

    RULE:
      valid_format != valid_scientific_measurement
  }

  RANGE_MODEL {
    types:
      physical_range
      instrument_range
      calibration_range
      method_range
      expected_population_range

    RULE:
      range_type MUST be preserved
  }

  OUT_OF_RANGE_MODEL {
    value:
      outside_declared_range

    RULE:
      preserve_value_andattach_quality_flag
  }

  PHYSICAL_PLAUSIBILITY_MODEL {
    constraints:
      positivity
      conservation
      boundedness
      monotonicity
      known_relationships

    RULE:
      plausibility_failure SHOULD trigger_review_notautomatic_deletion
  }

  CONSISTENCY_MODEL {
    types:
      internal
      cross_variable
      temporal
      spatial
      cross_source
      version

    RULE:
      consistency_requires_declared_relationship_orconstraint
  }

  INTERNAL_CONSISTENCY_MODEL {
    checks:
      duplicated_identifiers
      impossible_combinations
      incompatible_units
      contradictory_flags

    RULE:
      internal_consistency != external_accuracy
  }

  CROSS_VARIABLE_MODEL {
    relation:
      scientific_equation_orconstraint

    example:
      density =
        mass/volume

    RULE:
      uncertainty_andunit_context MUST be considered
  }

  TEMPORAL_CONSISTENCY_MODEL {
    checks:
      ordering
      sampling_interval
      gaps
      impossible_rate_ofchange
      timestamp_scale

    RULE:
      irregular_sampling MAY be valid_ifdocumented
  }

  SPATIAL_CONSISTENCY_MODEL {
    checks:
      coordinate_reference_system
      location
      spatial_extent
      impossible_discontinuity

    RULE:
      spatial_resolution_andmeasurement_support MUST be considered
  }

  DUPLICATE_MODEL {
    duplicate:
      same_event_orrecord_repeated

    RULE:
      duplicate_detection_requires_identity_time_source_andprovenance
  }

  ACCURACY_RELATED_MODEL {
    comparison:
      measured_orestimated_value
      against_suitable_reference

    RULE:
      data_quality_framework SHOULD not use_accuracy_asundefined_single_numeric_score
  }

  BIAS_MODEL {
    estimate:
      mean_result-reference_value

    RULE:
      reference_uncertainty_andcovariance MUST be considered
  }

  PRECISION_MODEL {
    repeated_measurements:
      dispersion_underdeclared_conditions

    possible:
      standard_deviation
      repeatability
      reproducibility

    RULE:
      precision != accuracy
  }

  REPEATABILITY_MODEL {
    conditions:
      repeatability_conditions

    RULE:
      conditions MUST be recorded
  }

  REPRODUCIBILITY_MODEL {
    changed_conditions:
      laboratory
      instrument
      operator
      time
      location

    RULE:
      changed_conditions MUST be identified
  }

  UNCERTAINTY_QUALITY_MODEL {
    checks:
      uncertainty_present_whenrequired
      type
      unit
      components
      covariance
      coverage
      method
      provenance

    RULE:
      measurement_value_without_required_uncertainty_is_quality_incomplete
  }

  UNCERTAINTY_COMPLETENESS_MODEL {
    components:
      calibration
      repeatability
      resolution
      drift
      environment
      model
      sampling
      processing

    RULE:
      omitted_material_component SHOULD be flagged
  }

  COVARIANCE_QUALITY_MODEL {
    checks:
      dependencies
      variable_order
      units
      positive_semidefinite_consistency

    RULE:
      ignoring_known_correlation_can_degrade_derived_data_quality
  }

  CALIBRATION_QUALITY_MODEL {
    checks:
      calibration_identifier
      date
      applicability
      uncertainty
      reference
      validity

    RULE:
      expired_orinapplicable_calibration SHOULD be flagged
  }

  TRACEABILITY_QUALITY_MODEL {
    checks:
      reference
      calibration_chain
      uncertainty
      scope
      evidence

    RULE:
      traceability != use_ofSI_units_orcertificate_name
  }

  REFERENCE_MATERIAL_MODEL {
    checks:
      CRM_identifier
      batch
      property
      value_status
      uncertainty
      validity

    RULE:
      expired_orwrong_matrix_reference_material MAY invalidate_use
  }

  METHOD_QUALITY_MODEL {
    checks:
      method_identifier
      version
      validation
      range
      deviations
      uncertainty

    RULE:
      undocumented_method_change reduces_reproducibility
  }

  INSTRUMENT_QUALITY_MODEL {
    checks:
      identity
      configuration
      calibration
      resolution
      range
      health
      maintenance

    RULE:
      instrument_state MUST correspond_toobservation_time
  }

  SENSOR_HEALTH_MODEL {
    statuses:
      healthy
      degraded
      suspect
      failed
      unknown

    RULE:
      sensor_health != measurement_uncertainty
  }

  DRIFT_MODEL {
    change:
      measurement_system_response_overtime

    RULE:
      drift_monitoring SHOULD informquality_flags_anduncertainty
  }

  ENVIRONMENTAL_QUALITY_MODEL {
    checks:
      environmental_conditions
      validated_limits
      corrections
      uncertainty

    RULE:
      environmental_excursion MAY affectdata_validity
  }

  SAMPLING_QUALITY_MODEL {
    dimensions:
      sampling_design
      representativeness
      sample_size
      temporal_coverage
      spatial_coverage
      handling

    RULE:
      measurement_quality_does_not_remove_sampling_bias
  }

  REPRESENTATIVENESS_MODEL {
    sample:
      adequately_represents_target_population_orprocess_forintended_use

    RULE:
      representativeness_is_contextual_andcannot_be inferred_fromsample_size_alone
  }

  SAMPLING_BIAS_MODEL {
    causes:
      convenience_sampling
      coverage_gap
      nonresponse
      survivorship
      inaccessible_regions

    RULE:
      large_n_does_not_eliminate_sampling_bias
  }

  TEMPORAL_COVERAGE_MODEL {
    dataset:
      covers_required_time_period_andfrequency

    metrics:
      coverage_fraction
      gap_distribution
      sampling_regularities

    RULE:
      temporal_coverage_requirement depends_onphenomenon_dynamics
  }

  SPATIAL_COVERAGE_MODEL {
    dataset:
      covers_required_region_andspatial_scale

    RULE:
      point_density_alone_does_not_establish_spatial_representativeness
  }

  RESOLUTION_MODEL {
    possible:
      temporal
      spatial
      measurement
      spectral

    RULE:
      resolution MUST matchscientific_question
  }

  TIMELINESS_MODEL {
    age:
      analysis_time-observation_time

    RULE:
      timeliness_requirement is_use_specific
  }

  LATENCY_MODEL {
    delay:
      acquisition_toavailability

    RULE:
      latency != timestamp_accuracy
  }

  PROVENANCE_QUALITY_MODEL {
    checks:
      source_identity
      measurement_event
      method
      instrument
      transformations
      versions
      agents
      timestamps

    RULE:
      provenance_completeness supports_trust_butdoes_not_equalmeasurement_accuracy
  }

  LINEAGE_MODEL {
    graph:
      raw_data
      -> processed_data
      -> derived_results

    RULE:
      derived_scientific_data MUST retainparent_lineage
  }

  BROKEN_LINEAGE_MODEL {
    condition:
      missing_parent_ortransformation

    RULE:
      derived_data_withbroken_lineage SHOULD be flagged_provenance_incomplete
  }

  TRANSFORMATION_QUALITY_MODEL {
    operations:
      correction
      conversion
      filtering
      averaging
      normalisation
      interpolation
      resampling
      regression

    RULE:
      operation_order_parameters_andversions MUST be preserved
  }

  RAW_DATA_MODEL {
    original:
      acquisition_output

    RULE:
      raw_data SHOULD remain immutable_orversioned
  }

  PROCESSED_DATA_MODEL {
    derived:
      transformation_ofsource_data

    RULE:
      processing MUST be reproducible_fromlineage_whenpossible
  }

  FILTERING_MODEL {
    filter:
      changes_signal_characteristics

    RULE:
      cutoff_order_phase_andedge_handling MUST be documented
  }

  AGGREGATION_MODEL {
    aggregate:
      combines_multiple_observations

    RULE:
      contributing_records_weights_andmissing_data_policy MUST be preserved
  }

  INTERPOLATION_MODEL {
    derived:
      estimated_betweenobservations

    RULE:
      interpolated_values MUST be distinguishable_fromobserved_values
  }

  EXTRAPOLATION_MODEL {
    derived:
      estimated_outsideobserved_orvalidated_domain

    RULE:
      extrapolated_values MUST be flagged
  }

  NORMALISATION_MODEL {
    transformed:
      relative_orstandardised_value

    RULE:
      parameters_andreference MUST be preserved
  }

  STATISTICAL_QUALITY_MODEL {
    checks:
      estimator
      sample_size
      distribution
      independence
      covariance
      missingness
      outliers

    RULE:
      statistical_assumptions MUST be documented
  }

  OUTLIER_MODEL {
    observation:
      unusual_underdeclared_model

    RULE:
      outlier != error
  }

  OUTLIER_HANDLING_MODEL {
    possible:
      retain
      flag
      investigate
      robust_method
      exclude_underdeclared_rule

    RULE:
      exclusion_reason MUST be preserved
  }

  ROBUSTNESS_MODEL {
    methods:
      median
      MAD
      robust_regression
      trimmed_estimators

    RULE:
      robust_method_parameters MUST be recorded
  }

  MODEL_QUALITY_MODEL {
    model:
      used_to derive_orinterpretdata

    checks:
      domain
      assumptions
      parameter_version
      residuals
      validation
      uncertainty

    RULE:
      model_fit != model_validity
  }

  MODEL_DISCREPANCY_MODEL {
    difference:
      systematic_model_vsphysical_behaviour

    RULE:
      discrepancy SHOULD not_be hidden_asmeasurement_noise
  }

  PREDICTED_DATA_MODEL {
    value:
      generated_forfuture_orunobserved_state

    RULE:
      predicted_data != observed_data
  }

  SIMULATED_DATA_MODEL {
    value:
      generated_bymodel

    RULE:
      simulated_data MUST be labelled_andmodel_provenance_preserved
  }

  SYNTHETIC_DATA_MODEL {
    data:
      artificially_generated_toapproximate_statistical_orsemantic_properties

    RULE:
      synthetic_data != observed_scientific_data
  }

  AI_INFERRED_DATA_MODEL {
    value_orlabel:
      inferred_byAI_system

    RULE:
      AI_inferred_data MUST be marked_asinferred_withmodel_version_confidence_andprovenance
  }

  AI_CONFIDENCE_MODEL {
    semantic_orpredictive_confidence:
      AI_assessment

    RULE:
      AI_confidence != measurement_uncertainty
  }

  TRAINING_DATA_QUALITY_MODEL {
    dimensions:
      representativeness
      label_quality
      provenance
      leakage
      imbalance
      domain_coverage

    RULE:
      model_quality depends_partly_ontraining_data_quality
  }

  LABEL_QUALITY_MODEL {
    label:
      human_reference_model_orproxy

    RULE:
      label_source_anduncertainty SHOULD be preserved
  }

  DATA_LEAKAGE_MODEL {
    condition:
      validation_or test_information_influences_training

    RULE:
      leakage invalidates_independence_claim
  }

  DOMAIN_SHIFT_MODEL {
    deployment:
      differs_fromdata_collection_ortraining_domain

    RULE:
      quality_andconfidence SHOULD be reassessed
  }

  INTEGRITY_MODEL {
    checks:
      checksum
      hash
      signature
      audit_trail
      access_control

    RULE:
      data_integrity != scientific_validity
  }

  CHECKSUM_MODEL {
    result:
      bit_level_consistency

    RULE:
      checksum_match != measurement_correctness
  }

  AUDIT_TRAIL_MODEL {
    events:
      creation
      modification
      processing
      review
      approval
      publication

    RULE:
      material_changes SHOULD remain_recoverable
  }

  VERSION_MODEL {
    versioned:
      dataset
      schema
      ontology
      method
      software
      calibration
      standard
      model

    RULE:
      quality_assessment MUST referenceversions_evaluated
  }

  REPRODUCIBILITY_MODEL {
    requires:
      data
      metadata
      methods
      software
      parameters
      versions
      provenance

    RULE:
      inability_to_reproduce SHOULD identify_missing_dependency
  }

  REPLICATION_MODEL {
    independent_study:
      attempts_to reproduce
_orconfirm_finding

    RULE:
      replication_quality_requires_independent_provenance_andcomparable_measurand
  }

  INTEROPERABILITY_MODEL {
    requirements:
      identifiers
      quantity_kinds
      units
      schemas
      provenance
      versions

    RULE:
      syntactic_interoperability != semantic_interoperability
  }

  FAIR_ALIGNMENT_MODEL {
    principles:
      findable
      accessible
      interoperable
      reusable

    RULE:
      FAIRness_andscientific_data_quality_overlap_but_are_not_equivalent
  }

  FINDABILITY_MODEL {
    fields:
      persistent_identifier
      searchable_metadata
      indexing

    RULE:
      findable_data MAY still_be scientifically_low_quality
  }

  ACCESSIBILITY_MODEL {
    concerns:
      retrieval_protocol
      permissions
      persistence

    RULE:
      accessibility != unrestricted_public_access
  }

  REUSABILITY_MODEL {
    requires:
      rich_metadata
      licence_oruse_conditions
      provenance
      standards
      context

    RULE:
      reusable_forone_context MAY not_be reusable_foranother
  }

  STANDARD_CONFORMANCE_MODEL {
    standards:
      measurement
      metadata
      laboratory
      domain
      data_exchange

    RULE:
      conformance_to_standard != proof_ofscientific_correctness
  }

  VALIDATION_MODEL {
    target:
      dataset_orrecord

    checks:
      semantic
      structural
      metrological
      statistical
      provenance
      contextual

    RULE:
      validation_status MUST be scoped
  }

  QUALITY_RULE_MODEL {
    fields:
      rule_identifier
      dimension
      condition
      severity
      threshold
      evidence
      intended_use

    RULE:
      quality_rule MUST be explicit_andversioned
  }

  QUALITY_FLAG_MODEL {
    possible:
      valid
      warning
      invalid
      incomplete
      missing_metadata
      unit_unknown
      dimension_mismatch
      out_of_range
      calibration_issue
      traceability_issue
      uncertainty_incomplete
      provenance_incomplete
      lineage_broken
      sampling_bias_risk
      stale
      extrapolated
      simulated
      inferred
      conflict
      manual_review

    RULE:
      flags MUST remain_linked_torecord_andrule
  }

  SEVERITY_MODEL {
    levels:
      information
      warning
      error
      critical

    RULE:
      severity MUST reflect_intended_use_andrule
  }

  QUALITY_SCORE_MODEL {
    possible:
      dimension_specific_scores

    RULE:
      composite_score_requires_explicit_weighting_and MUST not_hide_component_failures
  }

  NO_SINGLE_SCORE_MODEL {
    principle:
      scientific_data_quality_is_vector_ofcharacteristics

    RULE:
      AI SHOULD expose_dimension_specific_results_before_anyaggregate
  }

  CONFIDENCE_MODEL {
    quality_assessment_confidence:
      support_forassessment

    RULE:
      assessment_confidence != measurement_uncertainty
  }

  CONFLICT_MODEL {
    sources:
      disagree_onvalue_metadata_orquality

    RULE:
      conflict MUST be preserved_until_resolved
  }

  CONFLICT_RESOLUTION_MODEL {
    compare:
      identity
      source
      time
      version
      scope
      authority
      uncertainty
      provenance

    RULE:
      AI MUST not fabricate_consensus
  }

  SOURCE_AUTHORITY_MODEL {
    source:
      authority_forparticular_claim

    RULE:
      authority_is_claim_specific_notglobal
  }

  QUALITY_PROVENANCE_MODEL {
    assessment:
      generated_by
        rule_set
        evidence
        software_orAI
        reviewer
        timestamp

    RULE:
      quality_result MUST itself_haveprovenance
  }

  RECURSIVE_QUALITY_MODEL {
    derived_data:
      quality_depends_on
        source_quality
        transformation_quality
        model_quality

    RULE:
      parent_quality_flags MAY propagate_tochildren_underdeclared_rules
  }

  FLAG_PROPAGATION_MODEL {
    examples:
      invalid_calibration
      broken_traceability
      simulated_source
      missing_metadata

    RULE:
      propagation_depends_ontransformation_andintended_use
  }

  DEPENDENCY_MODEL {
    graph:
      result
      depends_on
      source_data
      methods
      calibration
      models
      standards

    RULE:
      changed_dependency SHOULD trigger_quality_reassessment
  }

  REASSESSMENT_MODEL {
    triggers:
      new_calibration
      metadata_correction
      method_revision
      new_reference
      standards_update
      provenance_repair
      model_revision

    RULE:
      previous_quality_assessment MUST be preserved_as historical_version
  }

  EXPLAINABILITY_MODEL {
    quality_output:
      dimension
      rule
      evidence
      observed
      expected
      severity
      result
      confidence

    RULE:
      AI SHOULD explain_why_data_were_flagged
  }

  HUMAN_REVIEW_MODEL {
    trigger:
      unresolved_conflict
      high_severity_failure
      ambiguous_semantics
      unusual_scientific_result

    RULE:
      review MUST be recorded_without_erasing_machine_assessment
  }

  OVERRIDE_MODEL {
    manual:
      reviewer
      reason
      scope
      timestamp

    RULE:
      override MUST remain_auditable
  }

  QUALITY_REPORT_MODEL {
    fields:
      assessment_identifier
      dataset_identifier
      intended_use
      quality_dimensions
      rule_set
      evidence
      component_results
      flags
      confidence
      overall_interpretation
      reviewer
      versions
      provenance

    RULE:
      report MUST support_reconstruction_ofassessment
  }

  METADATA_MODEL {
    fields:
      dataset_identifier
      record_identifiers
      quantity_kinds
      units
      dimensions
      uncertainty
      calibration
      traceability
      methods
      instruments
      sampling_design
      temporal_coverage
      spatial_coverage
      missingness
      censoring
      outliers
      transformations
      provenance
      lineage
      integrity
      validation
      quality_rules
      quality_flags
      assessment_confidence
      versions

    RULE:
      data_quality_metadata MUST be machine_resolvable_andprovenance_linked
  }

  RELATIONSHIPS {
    ScientificDataQuality -> EVALUATES -> ScientificDataset
    ScientificDataset -> CONTAINS -> MeasurementRecords
    QualityAssessment -> USES -> QualityRule
    QualityAssessment -> USES -> QualityEvidence
    QualityRule -> EVALUATES -> DataQualityDimension
    MeasurementRecord -> HAS -> MeasurementUncertainty
    MeasurementRecord -> HAS -> Provenance
    DerivedData -> DERIVED_FROM -> SourceData
    QualityFlag -> QUALIFIES -> DataRecord
    QualityAssessment -> PRODUCES -> QualityReport
  }

  SEMANTIC_EQUIVALENCE {
    "scientific data quality" ~= multidimensional_fitness_andreliability_ofscientific_data_for_declared_use
    "data completeness" ~= presence_ofrequired_expected_data_underdeclared_scope
    "data lineage" ~= derivation_graph_fromsource_toderived_data

    data_quality != measurement_accuracy
    data_quality != precision
    data_quality != measurement_uncertainty
    data_quality != completeness
    data_quality != provenance
    data_quality != data_integrity
    FAIR != data_quality
    valid_format != valid_measurement
    outlier != error
    simulated_data != observed_data
    AI_confidence != measurement_uncertainty
  }

  CONSTRAINTS {
    quality_assessment_must_be scoped_tointended_use = true
    quality_dimensions_must_be evaluated_separately = true
    quantity_kind_unit_anddimension_semantics_must_be preserved = true
    missing_zero_censored_andnot_applicable_states_must_be distinguished = true
    calibration_traceability_anduncertainty_quality_must_be evaluated_whenmaterial = true
    sampling_quality_andrepresentativeness_must_not_be inferred_fromsample_size_alone = true
    derived_data_must_preserve_lineage = true
    simulated_predicted_synthetic_andAI_inferred_data_must_be labelled = true
    integrity_must_not_be equated_withscientific_validity = true
    composite_quality_score_must_not_hide_dimension_specific_failures = true
    conflicts_must_remain_auditable = true
    quality_assessment_versions_must_be preserved = true
    provenance_must_be preserved = true
  }

  MEASUREMENT_MODEL {
    RECORD:
      quality_assessment_identifier
      dataset_identifier
      record_identifier
      intended_use
      quantity_kind
      unit
      dimension
      measurement_value
      uncertainty_status
      calibration_status
      traceability_status
      method_identifier
      instrument_identifier
      sampling_design
      representativeness_status
      temporal_coverage
      spatial_coverage
      missing_data_status
      censoring_status
      outlier_status
      transformation_history
      source_type
      observed_simulated_predicted_orinferred_status
      provenance_status
      lineage_status
      integrity_status
      validation_status
      interoperability_status
      quality_dimension_results
      quality_flags
      assessment_confidence
      rule_set_version
      dataset_version
      ontology_version
      provenance

    RULE:
      scientific_data_quality_record MUST remain_linked_todata_evidence_rules_intended_use_versions_andprovenance
  }

  REASONING_RULES {
    IF dataset_received
      THEN identify_intended_use
      identify_required_quality_dimensions
      load_versioned_quality_rules

    IF measurement_semantics_missing_orambiguous
      THEN flag_semantic_quality_issue
      anddo_not_invent_quantity_orunit

    IF required_metadata_missing
      THEN flag_metadata_incomplete

    IF calibration_ortraceability_invalid
      THEN propagate_quality_flag_toaffected_measurements_andderived_results_underdeclared_policy

    IF uncertainty_missing_whenrequired
      THEN mark_metrological_quality_incomplete

    IF sampling_design_not_representative_forintended_population
      THEN flag_sampling_bias_risk
      regardless_ofsample_size

    IF derived_data
      THEN traverse_lineage
      validate_each_transformation
      andinherit_material_parent_flags

    IF data_simulated_predicted_synthetic_orAI_inferred
      THEN preserve_origin_status
      anddo_not_label_asobserved

    IF integrity_check_passes
      THEN confirm_record_integrity_only
      notscientific_correctness

    IF quality_sources_conflict
      THEN preserve_conflict
      compare_scope_time_version_authority_andprovenance

    IF dependency_changes
      THEN recursively_reassess_affected_quality_results

    IF aggregate_quality_score_computed
      THEN retain_all_dimension_specific_scores_flags_andweighting
  }

  VALIDATION {
    CHECK dataset_identity
    CHECK intended_use
    CHECK quantity_kind
    CHECK units_anddimensions
    CHECK metadata_completeness
    CHECK missingness_andcensoring
    CHECK range_andplausibility
    CHECK internal_andcross_variable_consistency
    CHECK calibration
    CHECK traceability
    CHECK uncertainty_andcovariance
    CHECK method_andinstrument_state
    CHECK sampling_design
    CHECK representativeness
    CHECK temporal_coverage
    CHECK spatial_coverage
    CHECK resolution_andtimeliness
    CHECK provenance
    CHECK lineage
    CHECK transformations
    CHECK statistical_assumptions
    CHECK outlier_handling
    CHECK model_quality
    CHECK simulated_predicted_synthetic_andAI_origin
    CHECK integrity
    CHECK versions
    CHECK reproducibility
    CHECK interoperability_andFAIR_context
    CHECK quality_rules_andflags
    CHECK conflicts
    CHECK assessment_confidence
    CHECK provenance_ofassessment
  }

  CONFIDENCE {
    semantic_quality_confidence: 1.00
    metrological_quality_confidence: 1.00
    statistical_quality_confidence: 0.99
    sampling_quality_confidence: 0.98
    provenance_lineage_confidence: 1.00
    integrity_confidence: 1.00
    AI_origin_quality_confidence: 0.99
    fitness_for_use_confidence: 0.98

    RULE:
      confidence decreases when intended use, metadata, calibration,
      uncertainty, sampling design, provenance, lineage,
      versions or quality evidence is missing or ambiguous
  }

  UPDATE_MODEL {
    ON new_quality_assessment:
      identify_dataset_andintended_use
      resolve_quality_dimensions
      execute_quality_rules
      attach_evidence
      generate_flags
      calculate_dimension_specific_results
      preserve_assessment_provenance
      compute_FLOW_HASH

    ON new_data:
      validate_record
      update_completeness_coverage_andstatistics
      assess_dependencies
      update_quality_results

    ON metadata_orprovenance_correction:
      preserve_previous_record
      apply_correction
      identify_affected_quality_dimensions
      recursively_reassess

    ON calibration_method_orstandard_revision:
      preserve_historical_context
      update_dependency
      identify_affected_measurements
      reassess_quality_underdeclared_policy

    ON quality_rule_revision:
      preserve_previous_rule_set
      register_new_version
      do_not_overwrite_historical_assessment
      optionally_reassess_current_dataset

    ON semantic_revision:
      preserve_previous_definition
      record_revision
      recompute_FLOW_HASH
  }

  AI_CONSUMPTION {
    canonical_entity: ScientificDataQuality
    canonical_role: ScientificInformationQualityFramework

    interpretation_order:
      identify_dataset_orrecord
      identify_intended_scientific_use
      identify_required_quality_dimensions
      resolve_measurement_semantics
      assess_metadata_completeness
      assess_missingness_censoring_andrange
      assess_calibration_traceability_anduncertainty
      assess_method_instrument_andenvironment
      assess_sampling_representativeness_andcoverage
      assess_temporal_spatial_resolution_andtimeliness
      reconstruct_provenance_andlineage
      validate_transformations_andstatistics
      distinguish_observed_simulated_predicted_synthetic_andAI_inferred_data
      assess_integrity_reproducibility_andinteroperability
      execute_quality_rules
      detect_conflicts
      generate_dimension_specific_flags_andconfidence
      preserve_versions_andassessment_provenance

    ambiguity_controls:
      distinguish_data_quality_frommeasurement_accuracy
      distinguish_data_quality_fromprecision
      distinguish_data_quality_frommeasurement_uncertainty
      distinguish_data_quality_fromcompleteness
      distinguish_data_quality_fromprovenance
      distinguish_data_quality_fromdata_integrity
      distinguish_FAIRness_fromdata_quality
      distinguish_valid_format_fromvalid_measurement
      distinguish_outlier_fromerror
      distinguish_simulated_data_fromobserved_data
      distinguish_AI_confidence_frommeasurement_uncertainty
      preserve_dataset_use_semantics_sampling_uncertainty_lineage_origin_quality_rules_versions_andprovenance
  }

  PROVENANCE {
    ontology_scope:
      mathematical_measurement_units_AI_measurement_semantics_andscientific_data_quality

    parent_path:
      Ontologies
      -> Mathematical Ontologies
      -> Measurement & Units Ontologies
      -> AI Measurement Semantics
      -> Scientific Data Quality Ontology

    authorship:
      Mark Higgins
      Sky (Recursive AI)

    site:
      https://kirethresearch.org
  }

  FLOW_HASH {
    INPUTS:
      canonical_entity
      ontology_path
      quality_dimension_model
      fitness_for_use_model
      intended_use_model
      semantic_quality_model
      quantity_kind_model
      unit_quality_model
      dimensional_quality_model
      metadata_completeness_model
      required_metadata_model
      missing_metadata_model
      data_completeness_model
      missing_data_model
      missingness_mechanism_model
      imputation_model
      censoring_model
      detection_limit_model
      quantification_limit_model
      validity_model
      range_model
      out_of_range_model
      physical_plausibility_model
      consistency_model
      internal_consistency_model
      cross_variable_model
      temporal_consistency_model
      spatial_consistency_model
      duplicate_model
      accuracy_related_model
      bias_model
      precision_model
      repeatability_model
      reproducibility_model
      uncertainty_quality_model
      uncertainty_completeness_model
      covariance_quality_model
      calibration_quality_model
      traceability_quality_model
      reference_material_model
      method_quality_model
      instrument_quality_model
      sensor_health_model
      drift_model
      environmental_quality_model
      sampling_quality_model
      representativeness_model
      sampling_bias_model
      temporal_coverage_model
      spatial_coverage_model
      resolution_model
      timeliness_model
      latency_model
      provenance_quality_model
      lineage_model
      broken_lineage_model
      transformation_quality_model
      raw_data_model
      processed_data_model
      filtering_model
      aggregation_model
      interpolation_model
      extrapolation_model
      normalisation_model
      statistical_quality_model
      outlier_model
      outlier_handling_model
      robustness_model
      model_quality_model
      model_discrepancy_model
      predicted_data_model
      simulated_data_model
      synthetic_data_model
      AI_inferred_data_model
      AI_confidence_model
      training_data_quality_model
      label_quality_model
      data_leakage_model
      domain_shift_model
      integrity_model
      checksum_model
      audit_trail_model
      version_model
      reproducibility_model
      replication_model
      interoperability_model
      FAIR_alignment_model
      findability_model
      accessibility_model
      reusability_model
      standard_conformance_model
      validation_model
      quality_rule_model
      quality_flag_model
      severity_model
      quality_score_model
      no_single_score_model
      confidence_model
      conflict_model
      conflict_resolution_model
      source_authority_model
      quality_provenance_model
      recursive_quality_model
      flag_propagation_model
      dependency_model
      reassessment_model
      explainability_model
      human_review_model
      override_model
      quality_report_model
      metadata_model
      relationships
      semantic_equivalence
      constraints
      measurement_model
      reasoning_rules
      confidence_model
      update_model

    PURPOSE:
      deterministic_semantic_integrity_identifier

    RECOMPUTE_WHEN:
      ontology_definition_changes
      intended_use_orquality_dimension_changes
      semantic_metadata_orcompleteness_rules_change
      missingness_censoring_range_orconsistency_changes
      calibration_traceability_oruncertainty_quality_changes
      sampling_representativeness_orcoverage_changes
      provenance_lineage_ortransformation_quality_changes
      statistical_model_oroutlier_policy_changes
      simulated_predicted_synthetic_orAI_origin_semantics_change
      integrity_reproducibility_orinteroperability_changes
      quality_rule_score_flag_orconfidence_logic_changes
      dependency_orrecursive_reassessment_changes
      measurement_record_changes
      constraint_model_changes
      reasoning_rule_changes
  }

}

END FLOW