The challenge with traditional Recommendations
- Manual interpretation. Requirements are written in natural language. Engineers, test laboratories, and conformity assessment bodies must interpret and manually translate them into test procedures and acceptance criteria. Different interpretations lead to inconsistent results.
- Siloed Recommendations. Each International Recommendation is self-contained. There is no shared vocabulary or machine-readable structure connecting related Recommendations. Cross-references are implicit and must be tracked manually.
- Spreadsheet-based test reports. Test laboratories fill out test report forms using spreadsheets or paper. Formulas are copied by hand, lookup tables are recreated, and transcription errors are common. Verification of calculations is labor-intensive.
- Delayed adoption. When a new Recommendation is published or an existing one is amended, the time between publication and practical implementation by member states is measured in years, partly because of the manual work required to interpret and operationalize the changes.
The four SMART principles
The acronym SMART captures what the programme delivers:
- Standards. OIML International Recommendations transformed from static PDF documents into structured digital artifacts.
- Machine-Actionable. Built-in computation engines perform metrological calculations directly in the browser. No manual lookup tables or spreadsheet formulas needed.
- Readable. Machine-readable requirements, tests, and terminology in open formats. Instant full-text search across all provisions with automatic cross-references.
- Transferrable. A shared OWL ontology enables semantic queries, reasoning, and cross-Recommendation interoperability across all OIML recommendations.
Before and after SMART
| Aspect | Before SMART | After SMART |
|---|---|---|
| Authoring | AsciiDoc + PDF | AsciiDoc + YAML model + PDF |
| Requirements | Natural-language paragraphs | Typed requirements classes with structured criteria |
| Acceptance criteria | Tables and formulas in prose | Executable expressions and lookups |
| Test reports | Spreadsheets, manual calculation | Computation-powered forms, auto-calculated |
| Cross-references | Implicit, manually tracked | Explicit, machine-readable URN links |
| Terminology | Glossaries at the end of the document | OWL ontology shared across Recommendations |
| Adoption cycle | Years | Weeks |
The result: a Recommendation that is not just read but executed.
Read the architecture chapters for the technical details.