Skip to content

INDUSTRY FIELD GUIDE

Iron ore grade-basis records for AI evaluation

An ore sample and a concentrate report can use similar field names while describing different materials.

By Fiund · Updated 2026-11-05

NAICS 212210 · Iron Ore Mining

An ore sample and a concentrate report can use similar field names while describing different materials. USGS publishes iron ore statistics and geochemical data-quality guidance. A mine can demonstrate a basis check with artificial assays and product labels. USGS iron ore data release · USGS geochemical data quality primer

Create two practice samples and label material, measurement basis, units and report revision. Put a concentrate figure next to an ore figure with an unspecified basis. Ask which comparisons are supported and which require a reviewer. Add a corrected sample association that does not change the measurement.

Keep actual assays, grades, product specifications, locations, client contracts, operating and processing information private. Use independently written mock records and no extraction or process task.

Describe a grade-basis test

Original Fiund exercise

Material identity and reporting basis must both survive an assay join

The invented packet contains an ore specimen, a concentrate specimen and a corrected sample association. No actual grade, processing relationship or ore valuation is represented.

Fictional records supplied to the evaluator
RecordSourceWhat it records
IO1Practice assay report O-1Specimen X: material ore; field F=24 percent; reporting basis unspecified.
IO2Practice assay report C-1Specimen Y: material concentrate; F=61 percent; basis dry.
IO3Association correctionO-1's specimen identifier is X2, replacing X. The correction changes no value, material label or reporting basis.

Question for the model

What does the correction establish, and is a direct grade comparison or enrichment calculation supported between X2 and Y?

Show the reference answer and checks

IO3 changes O-1's specimen association to X2 while retaining its ore label, 24 percent value and unspecified basis. IO2 concerns a concentrate on a dry basis. The packet supplies neither a common basis nor a documented feed-to-product relationship. A direct like-for-like grade comparison or enrichment calculation would therefore add assumptions beyond the records.

  • Treating a sample-identifier correction as a newly measured assay.
  • Calculating a processing improvement from ore and concentrate values with unmatched basis and no process lineage.

Fields to preserve

Material class
Separates an ore specimen from a concentrate specimen.
Reporting basis
Retains the missing basis rather than silently assuming dry mass.
Corrected identifier and prior alias
Maintains traceability through the association change.
Feed-to-product relationship status
Makes a missing processing link explicit.

Test another case

Give X2 a verified dry-basis statement but no connection to Y. The model may resolve comparability of the stated measure while still declining to claim that Y was produced from X2.

Keep actual grades, reserves, product specifications, locations, contracts and processing information private. All records here are independently authored fiction.

Download this fictional example and answer key (JSON)

Discuss a contribution

Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.

Primary sources and further reading