Working repository for the SMHE / SectionHub Material Handling Taxonomy Standard and the public, deterministic Taxonomy Explorer application built on it.
| Path | What it is |
|---|---|
| SMHE-SectionHub-Taxonomy-Standard-DRAFT.md | The full v1.0 draft standard: identity layers, namespaces, code dictionary, decision rules, worked examples, Explorer specification, governance & adoption strategy (Parts 0–15) |
| dictionaries/ | Seed dictionary CSVs (namespaces, section shapes, component roles, assembly types, product families, configuration groups) matching the draft's Part 3 tables, pinned to SNAP-1.0.0 |
| standards/ | Placeholder homes for the split standards documents (GSID 2D, MH Taxonomy, Code Dictionary) per the draft's Part 15 outlines |
| CONTRIBUTING.md | How to propose codes and changes: evidence gates, proposal workflow, determinism & AI policy |
| GOVERNANCE.md | Roles, decision rights, release train, industry participation, amendment process |
| CODE_OF_CONDUCT.md | Contributor Covenant 2.1 plus project-specific standards for a vendor-neutral standards effort |
| ROADMAP.md | Seven-phase plan from foundation to formal governance |
| docs/repository-structure.md | Target repository layout for the data layer and future Explorer application, with rationale and migration plan |
| explorer/ | Explorer (Phases 1–2) — pure static, deterministic taxonomy navigation demo with rule traceability (GitHub Pages-ready; see explorer/README.md) |
The Explorer is deterministic and traceable by construction: every classification is a pure function of your answers plus versioned dictionaries (explorer/data/) and a machine-readable rule table (explorer/rules/question_nodes.csv). Each result shows a Decision trace — the exact rule IDs applied, the dictionary rows used, the reason, and the standard reference — and embeds it in the JSON output. When required input is missing, the engine refuses to guess (warning W-201, status INDETERMINATE) and states what information is needed. The current dictionaries are an honestly-labeled demo subset (SNAP-0.1.0-DEMO), not the final standard; known deviations are tracked as an open governance item. Dictionary and rule changes go through the proposal process in CONTRIBUTING.md — never through the app.
Identity is split into separate but compatible layers: canonical 2D section geometry (SEC: shape codes → CanonicalGeometryID → GSID), domain-scoped product classification (FAM: / ASM: / ROL: codes), content-addressed ConfigurationIDs for components and assemblies, a zero-authority naming layer (manufacturer parts and aliases), an evidence layer (tests and certifications), and immutable governance snapshots. GSID identifies canonical 2D cross-section geometry only — assemblies, products, and SKUs are classified by family, assembly type, component roles, and configuration identity, never by GSID.
DRAFT for senior-engineer review, dated 2026-07-02. Nothing here is a released snapshot; all example identifiers in the draft are illustrative.
The Phase 1 TaxonomyExplorer MVP is a static deterministic browser app.
It demonstrates:
- Section classification
- Component role classification
- Assembly classification
- Suggested code output
- JSON output
- refusal-to-guess behavior for indeterminate cases
Live demo: