Assessment Domain Ontology

FLOW ALGEBRA
TITLE: Assessment Domain Ontology
VERSION: 1.0
SOURCE: https://kirethresearch.org/booksai-core/ontologies/domain-ontologies/education-domains/assessment-domain-ontology.html
STATUS: Domain-based draft; source page content not independently verified.
AUTHORS: Mark Higgins — Sky (Recursive AI)

DOMAIN:
  id: education.assessment
  parent: education
  purpose: Model educational assessment, evidence of learning, interpretation, feedback and decision-making with explicit validity and uncertainty.

TYPED_ENTITIES:
  Learner: Person
  Educator: Person
  Assessor: Person | System
  Assessment: Activity
  AssessmentInstrument: Instrument
  AssessmentItem: Question | Task | Observation
  LearningOutcome: CompetencyStatement
  Competency: Knowledge | Skill | Disposition
  Criterion: EvaluationRule
  Rubric: CriterionSet
  Response: LearnerEvidence
  Score: NumericMeasure
  Grade: OrdinalCategory
  Feedback: Communication
  Accommodation: AccessibilityAdjustment
  AssessmentDecision: Decision
  EvidenceRecord: ProvenanceRecord

RELATIONSHIPS:
  Assessment -> MEASURES -> LearningOutcome
  Assessment -> USES -> AssessmentInstrument
  AssessmentInstrument -> CONTAINS -> AssessmentItem
  AssessmentItem -> ELICITS -> Response
  Response -> PROVIDES_EVIDENCE_FOR -> Competency
  Rubric -> DEFINES -> Criterion
  Assessor -> EVALUATES -> Response
  Score -> DERIVED_FROM -> Response
  Grade -> DERIVED_FROM -> Score | Rubric
  Feedback -> REFERENCES -> EvidenceRecord
  Accommodation -> MODIFIES_ACCESS_TO -> Assessment
  AssessmentDecision -> SUPPORTED_BY -> EvidenceRecord

ASSESSMENT_TYPES:
  formative: Guides ongoing learning and feedback.
  summative: Summarises achievement at a defined point.
  diagnostic: Identifies existing knowledge or learning needs.
  criterion_referenced: Compares evidence against defined criteria.
  norm_referenced: Compares results against a reference population.
  authentic: Evaluates performance in a meaningful applied task.
  self_assessment: Learner evaluates their own work.
  peer_assessment: Learners evaluate peers' work under specified criteria.

INFERENCE_RULES:
  R1: IF assessment MEASURES outcome AND response satisfies criterion THEN infer evidence_of_attainment(outcome), qualified by validity and reliability.
  R2: IF response fails a criterion THEN infer possible_learning_gap(criterion), NOT definitive inability.
  R3: IF multiple independent valid evidence records support a competency THEN strengthen confidence in competency attainment.
  R4: IF assessment instrument lacks accessibility for a learner THEN flag potential construct-irrelevant barrier; do not infer low competency directly.
  R5: IF score or grade is derived THEN retain instrument, rubric version, date, assessor and transformation provenance.
  R6: IF assessment conditions or standards change THEN require reassessment of cross-cohort comparability.
  R7: IF formative evidence identifies a gap THEN recommend targeted feedback and further assessment, not automatic high-stakes action.

CONSTRAINTS:
  - Do not equate test scores with intelligence, personal worth or fixed ability.
  - Do not infer causality from a single assessment result.
  - Distinguish observed performance from latent competency.
  - Preserve context, accommodations, uncertainty and relevant population.
  - Avoid demographic proxies and unsupported fairness claims.
  - Respect learner privacy and lawful data handling.

CONFIDENCE:
  range: [0.0, 1.0]
  interpretation: Confidence in an evidence-supported proposition, not an objective probability without calibration.
  factors: [instrument_validity, reliability, evidence_quantity, assessor_consistency, context_fit, recency]
  rule: Missing validation evidence must reduce certainty, not be silently treated as high confidence.

SEMANTIC_EQUIVALENCE:
  assessment: [educational_evaluation, learning_assessment]
  rubric: [scoring_guide, evaluation_criteria_matrix]
  formative_assessment: [assessment_for_learning]
  summative_assessment: [assessment_of_learning]
  note: Synonyms are context-dependent; grading, testing and assessment are not universally identical.

PROVENANCE:
  required: [source, author_or_assessor, timestamp, instrument_version, rubric_version, jurisdiction_or_institution, methodology]
  rule: Every assessment claim should trace to evidence and applicable standards.

FLOW_HASH:
  algorithm: SHA-256
  input: Canonical UTF-8 serialisation of ontology identifier, version, entities, relationships, rules and constraints.
  value: COMPUTE_AT_PUBLICATION

UPDATE_MODEL:
  triggers: [new_assessment_evidence, rubric_revision, standards_revision, validation_study, accessibility_change]
  actions: [validate, version, recompute_inferences, recalculate_confidence, recompute_flow_hash, log_changes]
  policy: Preserve previous versions and avoid retroactively changing historical results without an audit trail.

AI_CONSUMPTION:
  parse_mode: typed_semantic_graph
  expected_outputs: [competency_evidence_map, assessment_alignment_report, feedback_suggestions, uncertainty_report]
  safeguards: [human_review_for_high_stakes_decisions, provenance_required, privacy_preserving, no_unjustified_learner_profiling]
  failure_mode: If evidence is insufficient, output UNKNOWN with reasons rather than fabricate a grade or conclusion.

END FLOW ALGEBRA