Skip to content

Technical Evidence Overview

Companion evidence pack for the Product Holon Architecture. These pages retain observations, validation evidence, examples, ADRs, open issues, and references.

Technical Evidence and Reference Implementation

Version: 1.3.0
Date: 17 August 2026
Status: Working Draft — companion to GS1_Product_Holon_Architecture_v2.1.0.md
Authority notice: The Product Holon is a proposed architectural construct. The evidence pack distinguishes approved GS1 capabilities, source-derived semantic artefacts, reference implementations, candidate mappings and proposed design choices.

Document control

Purpose

This document preserves the technical evidence, implementation detail and decision history underlying the core GS1 Product Holon Architecture. It replaces the chronological research-dossier structure of gs1-gdsn-holon-v1.31.0.md with an evidence-oriented structure. Superseded conclusions are not repeated as current recommendations; they are retained through Architecture Decision Records and explicit correction notes.

Relationship to the core architecture

The core architecture defines the proposed model, phases, components, governance and adoption roadmap. This companion provides:

  • the source inventory and verification boundaries;
  • direct observations from the supplied GDSN, GPC, SHACL and mapping artefacts;
  • the K10X use case that motivated the initial investigation;
  • worked Product Holons and SHACL examples;
  • the ontology and shape-generation pipeline;
  • mapping-candidate generation and SSSOM curation evidence;
  • OWL, DL-safe SWRL and SPARQL design exhibits;
  • validation and reconciliation results;
  • Architecture Decision Records;
  • limitations, unresolved questions and references;
  • the HTML Semantic Proof and Verifiable Product Statement reference profile, including a source-authority example, HTML scripts, credential structure and acceptance tests.

Evidence labels

Label Meaning in this pack
Observed Directly present in or computed from a supplied artefact.
Demonstrated Executed or validated in the source investigation or reference implementation.
Corroborated candidate Supported by more than one source or method but still awaiting formal governance review.
Proposed An architectural or implementation choice not established as an existing GS1 requirement.
Superseded A historical conclusion replaced by later evidence.
Unresolved The available sources do not establish the answer.

Editorial policy

The evidence pack uses an institutional technical voice. First-person session language, conversational corrections and version annotations in headings have been removed. The evidence itself remains traceable to the source document and named files.

Evidence Guide

Executive evidence summary

The source investigation began as an analysis of a K10X “Layer 1 — AI Data Interpretation” explainer and the role of GS1 Web Vocabulary in AI-enabled shopping. Direct inspection of supplied GDSN and GPC semantic artefacts broadened the work into a product-centred architecture. The evidence showed that:

  • GPC provides category scope through Bricks, Attribute Types and controlled Attribute Values;
  • GDSN provides deep product semantics, structured records, code lists and recipient-oriented business data;
  • Web Vocabulary and schema.org provide the public and agent-facing publication layer for a bounded subset;
  • a product-specific graph can be constructed by resolving a GTIN to a Brick and selecting only populated, applicable facts;
  • Brick-scoped SHACL shapes can be generated mechanically and validated with deterministic tooling;
  • the current gs1-gdsn-holon implementation independently converged on the same Brick-scoped generation and validation pattern;
  • real generated GDSN CURIEs are opaque, so name matching cannot be implemented as a simple namespace substitution;
  • lexical tools can generate useful candidate mappings but cannot approve them;
  • SSSOM provides an appropriate source of truth for candidate, reviewed and rejected mappings;
  • direct OWL equivalence is only sound for semantically and structurally compatible terms;
  • conditional or structural mappings require DL-safe rules or versioned transforms;
  • SPARQL remains an export/projection mechanism outside the ontology;
  • Product and Offer data require separate authority and provenance;
  • an LLM should retrieve and narrate validated facts rather than execute final rule logic.

The completed LOOM/BioPortal run produced 28,101 unique terms, 237 unique cross-ontology candidate pairs and 474 directional skos:closeMatch records. GDSN-to-Web Vocabulary candidates dominated, while GPC-to-Web Vocabulary candidates were sparse. Reconciliation with specific terms produced useful corroboration and several corrections, including the distinction between a flat GDSN allergen-statement attribute and the separate structured allergen-code chain.

This revision adds the publication proof required to make the architecture independently consumable. The proposed HTML Semantic Proof exposes a canonical GDSN/GPC-aligned Product Holon and a schema.org/GS1 Web Vocabulary discovery projection, connects them to source authority and SHACL evidence, and links a W3C Verifiable Product Statement that binds an authorised issuer to selected statements or integrity-protected resources. The credential is treated as proof of authorship, integrity and current status, not as automatic proof of factual truth.