Changelog

Changelog

v0.2.0 — 2026-06-30

Literature Integration & IOA-ORM Reframing.

Two major iterations (7 and 8) that significantly strengthen the theoretical grounding and contribution positioning.

Iteration 7: Literature Integration

Triggered by a collaborator’s literature review covering Fenton (2015), Joughin (2010), Sotiriadou (2020), and Bloom’s Taxonomy. Systematic gap analysis identified 10 gaps (1 critical, 4 moderate, 5 minor). All resolved.

New Schema Constructs:

• BloomLevel enum — 6 cognitive levels (remember → create)
• cognitiveLevel on EvidenceTarget — Bloom’s Taxonomy classification
• integrated_practice — new evidence dimension for multi-dimensional competence
• CandidateBriefing — candidate-facing exam information for transparency
• ConversationalStylePolicy — AI examiner tone and rapport-building
• FormativeFeedbackPolicy — learning-oriented feedback in formative mode
• promptingPrinciples on FollowUpPolicy — neutrality, consistency, transparency, reflexivity
• cognitiveEscalationStrategy on FollowUpPolicy — how follow-ups escalate cognitive depth
• bookPolicy on ExamMetadata — open/closed/restricted book configuration
• inter_item_consistency and intra_rater_reliability — new ValidityClaim types
• moderatorTraining on CalibrationProfile — human moderator training requirements
• isPractice and anxietyMitigation on warmup nodes — anxiety management as design concern
• ResourceReference utility type

New Theoretical Grounding:

• Bloom (1956) — cognitive level classification for evidence targets
• Sotiriadou et al. (2020) — interactive oral assessment definition
• Fenton (2015) — oral assessment as conversation, not interrogation
• Joughin (2010) — three-way classification, practical guidance
• Nguyen et al. (2023) — AI capability vs Bloom’s levels
• Huxham et al. (2012) — student anxiety in oral assessments

New Design Principles:

• P9: Cognitive Depth Is a Design Parameter
• P10: Transparency Builds Trust

Iteration 8: IOA-ORM Reframing

Triggered by an external framing review memo recommending repositioning from “Exam Runtime IR” to “IOA Ontology and Reference Model.”

Primary Name Change:

• “Exam Runtime IR” → “Interactive Oral Assessment Ontology and Reference Model” (IOA-ORM)

4-Layer Artifact Model:

1. Domain Ontology — shared vocabulary and semantics
2. Reference Model — reusable system abstraction
3. Executable Specification — machine-readable schema and validation
4. Intermediate Representation — engineering pipeline role

Key Changes:

• Formal definition paragraph (memo §7)
• DSR contribution mapping table (Constructs/Models/Methods/Instantiations)
• “ontology-grounded” qualification (not over-claiming full semantic-web)
• 5 new glossary terms (IOA-ORM, IOA Domain Ontology, IOA Reference Model, IOA Executable Specification, IR)
• Design Principle P1: “The IR Is the Contract” → “The Specification Is the Contract”
• 100+ prose replacements: “the IR” → “the specification” across all spec docs
• Schema type names preserved (ExamRuntimePackage, etc.)
• Implementation-agnostic framing: IOA need ≠ AI specifically

Deferred:

• Directory rename (exam-runtime-ir-spec/ → ioa-spec/)
• Repo rename (ExamRumtimeIR-DSR)
• Schema type name changes (never — published contract)

Documentation

• 8 design iterations documented in design-iterations/
• References expanded from 4 to 12+ oral assessment papers
• Glossary expanded with IOA-ORM terminology
• All spec documents updated with version information and revision history

---

v0.1.0 — 2026-05-06

Initial release.

Specification (15 documents)

• 00-overview.md — Purpose, goals, theoretical grounding, architecture
• 01-concepts.md — Theoretical foundations, glossary, domain entities
• 02-schema.md — TypeScript interfaces for 26 core object sections
• 03-runtime-semantics.md — State machine, transition rules, policy evaluation
• 04-agent-boundary.md — Allowed/forbidden actions, guardrail enforcement
• 05-event-protocol.md — Event types, payloads, delivery guarantees
• 06-evidence-ledger.md — Signal lifecycle, ledger schema, marking integration
• 07-pipecat-adapter.md — Compilation rules, FlowManager mapping, limitations
• 08-validation-rules.md — Compile-time validation (117 rules across 10 categories)
• 09-versioning.md — Version scheme, migration, compatibility
• 10-examples.md — 5 worked examples across Joughin’s dimensions
• 11-migration-plan.md — Incremental migration from flowJson (5 phases)
• 12-testing-strategy.md — Unit, integration, psychometric testing (25+ test cases)
• 13-open-questions.md — 21 unresolved design decisions
• 14-design-alternatives.md — 12 QOC design decisions with evaluation matrices

Design Iterations (5 phases)

• Iteration 1: 5 expert reviews of initial specification
• Iteration 2: 5 implementation changelogs (18 accepted, 14 rejected, 6 modified)
• Iteration 3: Consistency audit and reconciliation report
• Iteration 4: DSR contribution self-assessment (Hevner 7 Guidelines)
• Iteration 5: 12 design space deliberations (QOC methodology)

Knowledge Base

• 4 core oral assessment papers (Joughin 1998, Akimov & Malin 2020, Bayley et al. 2024, Fenton 2025)
• 12 IS/AI literature papers (agent systems, AI safety, audit, workflow)

DSR Review (Iteration 6)

• Full 4-pass DSR review (Pass 1: IRP/FRP, Pass 2: Solution-Problem, Pass 3: Evaluation, Pass 4: Contribution)
• Publication path assessment (DESRIST 75%, MISQ 40%)
• Interactive design space visualization (HTML)

Website

• Astro Starlight spec site with W3C-style normative language highlighting
• 18 spec pages, glossary, references, changelog
• GitHub Pages deployment