Separate reported test methods from assumed equivalent methods

Similar method names do not prove equivalence

Preserve what each supplier reported so the reviewer can assess comparability without inheriting an assumption introduced during extraction.

Our public material-register example shows a synthetic report beside downloadable output. Use it to inspect how values and source references travel together. The self-service ParseMill trial is for invoice PDFs; supplier test reports and engineering review need a separate fit discussion.

A shared test name such as density does not establish that methods produce directly comparable results. Sample preparation, conditions, measured quantity and reporting basis may differ. The register should retain those distinctions so a qualified reviewer can assess comparability.

Use a method-string register

The methods are invented labels, not real standards or claims about their technical content. The fixture’s purpose is to preserve differences that a later review may find material. A search-friendly short label can be added separately without replacing the full source string.

W3C: RDF Data Cube Vocabulary treats qualifying attributes separately from measured values. A reported method is useful context to retain alongside an observation rather than hiding it in a general document note.

Prepare questions for qualified review

1. Capture the exact method string and source location for each observation or verified range. Include revision, year and modification language.
2. Record source-stated preparation, conditions and reporting basis in separate fields where they affect comparison. Mark unstated information explicitly.
3. Ask whether the methods measure the same quantity under comparable conditions and whether a governing procedure accepts the proposed mapping. Retain the reviewer’s evidence and scope.
4. Keep any approved equivalence mapping versioned and separate from raw labels. Revisit it when a method revision or source qualification changes.

NIST: Metrological traceability distinguishes measurement-result traceability from merely having a report or instrument reference. Likewise, a method label alone is not proof of broader technical validity or comparability.

Avoid normalization that invents agreement

Removing revision suffixes can make different procedures look identical. Deleting the word modified can erase a limitation central to interpretation. Keep such text in the raw register even if the destination requires a separate standardized category.

If the source cites only a partial method name, do not complete it from a familiar standard or a neighboring row. Record the incomplete label and request the missing detail when the intended comparison depends on it, identifying the exact observation affected.

Check the evidence remains available

For the fixture, revisions 2 and 3 remain distinct and the modified preparation stays visible. S4 remains unstated. A reviewer can then decide what comparison is justified without reconstructing details lost during cleanup.

For a supplier-report scope discussion, bring exact method labels, revisions and modifications that your register must preserve. Agree the extraction and manual-review boundary before a pilot. Method equivalence remains a qualified decision outside the invoice product.