Changelog (AIPROD)
[0.1.6-closure] - Ontology Realignment Closure & Release Readiness (2020-07-23)
Added
- Release-readiness and migration/deprecation guidance.
- SHA-256 release integrity manifest for the AIPS 0.1 artifact set.
- Repository-wide static documentation, sidebar/front-matter, package-lock and release-manifest guards.
- Final convergence checks for interactive tools, machine profiles and cross-artifact lifecycle bundles.
Release posture
The ontology realignment is repository-conformant and release-candidate ready. Full Docusaurus production build and executable SHACL remain environment-dependent release checks and should be run in connected CI before publication.
Principle
AIPS release readiness requires both semantic conformance and artifact integrity; a successful individual schema validation is not sufficient evidence of a coherent release.
[0.1.5-covp] - Consumer Feedback, Value Realization & COVP Alignment (2020-07-22)
Added
- Consumption Observation & Value Plane (COVP) documentation and machine-readable profile.
- Version-bound usage telemetry, consumer feedback, contextual fitness and value-realization evidence.
- Attribution confidence for value claims and explicit fitness currentness.
- Marketplace projection of governed COVP summaries without ontology reclassification.
- Continuous Assurance routing from COVP evidence into controlled PDEP evolution.
Principle
Consumption evidence can change what should be improved, governed or re-evaluated; it does not by itself change what the Product intrinsically is.
[0.1.4-marketplace] - Marketplace, Discovery & Consumption Alignment (2020-07-22)
Added
- Explicit
productFamily=AIProductin the AIPROD publish projection. - Marketplace listing rules that preserve AI Product / Model Product / AI-enabled Product boundaries.
- Trust-aware current listing eligibility and historical-version semantics.
- Discovery/acquisition/consumption separation and governed consumption-port projection.
Principle
The marketplace projects immutable Product semantics and current trust evidence; it does not determine AI Product Family by presentation or marketing labels.
[0.1.3-governance] - Governance & Risk Ontology Alignment (2020-07-22)
Added
- Evidence-backed
RiskDriverAssessment. - Version-specific
GovernanceObligationProfile. - Controlled topology scope: Standalone, DomainCoupled, CrossDomainFederated, EnterpriseOrchestrated.
- Machine-checkable risk floors for autonomous, physical, closed-loop and enterprise-orchestrated action.
- AIPROD exposure of governance obligation profiles.
Changed
- Removed legacy
PrimaryArchetype/ secondary-capability stacking as the normative governance computation model. - R2+ now requires mechanism-neutral change/quality monitoring rather than universal model-drift monitoring.
- R3+ requires intervention/override and immutable audit; R4 requires executive escalation.
- Derived AI Product governance now explicitly includes dependency/currentness obligations where material.
This file records the release history of the AI Product Semantic Specification (AIPROD).
[0.1.2-machine] - Core/AIPROD Machine-readable Alignment (2020-07-21)
Added
- Machine-readable
AIQualificationEvidenceandAIPrimitiveEvidence. - Version-specific qualification status:
CURRENT,STALE,REVALIDATION_REQUIRED,INVALID. - Machine-readable APT with
PrimaryandDerived. SingleUpstreamandCompositederivation topology.- Machine-readable IC / CS / PU / MO / IE / AR / EM vocabularies.
ModelProductandAIPrimitiveboundary semantics.- Conformance examples for Primary, Derived, Composite-Derived, symbolic AI, Model Product boundary, and qualification invalidation/currentness.
- Repository validation helper at
scripts/validate-aips-0.1.py.
Changed
- Core JSON Schema, JSON-LD, RDF/Turtle and SHACL now implement the ontology-aligned AI Product Version model.
- AIPROD JSON Schema, JSON-LD, RDF/Turtle and SHACL now expose AQE, APT and orthogonal ontology dimensions.
- Marketplace publish projection now exposes APT, qualification currentness, IC, IE, AR and EM rather than relying on a single primary archetype.
PrimaryArchetypecompatibility terms remain available but are deprecated as umbrella classifications.
Boundary Rule
A ModelProduct is intentionally expected not to validate as an AIProductVersion merely because it is a governed, productized learned model.
[0.1.1-conceptual] — AI Product Ontology Realignment (2020-07-21)
Added
- AI Product Qualification & Evidence (AQE) as a version-specific semantic construct.
- AI Product Type (APT) with
PrimaryandDerivedintrinsic classifications. - Composite derivation as a topology of Derived AI Products rather than a peer Product Type.
- Explicit Intelligence Capability (IC) taxonomy.
- Qualification currentness and downstream intrinsic-classification impact semantics.
- Clear distinction between Model Product, AI Primitive, and AI Product.
Changed
- Replaced the overloaded Capability Type model with orthogonal ontology dimensions.
- Generalized lineage beyond model/training provenance to include symbolic, agentic, and non-model AI Primitives.
- Clarified Product identity vs immutable Product-version/AIPROD semantics.
- Updated composability so orchestration does not automatically create a Derived AI Product.
Compatibility Note
This tranche established the conceptual target. Core/AIPROD machine-readable alignment is delivered by 0.1.2-machine; AIPDS mechanism-neutral generalization is delivered by the subsequent Stage 4 tranche, while AIPS-DPP remains scheduled for ontology-aware generalization.
[0.1.0] — First Draft Release (2025-09-19)
Added
- Initial AI Product / AI Asset distinction.
- Initial AIPROD sections covering identity, intent, capability, I/O, lineage, governance, quality, observability, interoperability, economics, discoverability, lifecycle, explainability, composability, prohibited uses, extension, examples, and glossary.
Historical Note
The original Capability Type concept mixed Product Type, intelligence capability, agency, and intelligence mechanism. It is superseded by the ontology-aligned APT/IC/AR/MO/EM/IE model documented in this release.