INDUSTRY FIELD GUIDE
Industrial sand grade and test records for AI
A test summary can look complete while referring to a different product grade than the one requested.
NAICS 212322 · Industrial Sand Mining
A test summary can look complete while referring to a different product grade than the one requested. USGS distinguishes industrial sand from other aggregates in its commodity summaries. A producer can use dummy grades to show how a model should check product, method and documented scope. USGS industrial sand and gravel summary
Write a fictional request, a verified test for one product and a report for a lookalike grade. Give a later revision a change in the test basis. Ask which report answers the request and what evidence the other lacks. Use artificial values and no buyer acceptance decision.
Exclude actual product and processing details, quality thresholds, proprietary test results, clients, prices and locations. Create dummy product names and independently written references.
Describe a test matching benchmark
Original Fiund exercise
Choose the report that matches both the grade and the requested basis
An invented industrial-sand request names a grade, measure and basis. Similar product labels and a newer report test whether the assistant checks scope before quoting a result.
| Record | Source | What it records |
|---|---|---|
| IS1 | Fictional request R-12 | Report measure V for grade Sand-A on basis B1. |
| IS2 | Reviewed report TEST-4 | Grade Sand-A; measure V=16 practice units; basis B1; revision 1. |
| IS3 | Other reports | TEST-5: grade Sand-AX, V=18 on B1. TEST-4 revision 2: Sand-A, V=20 on basis B2; no B1 conversion rule supplied. |
Question for the model
Which supplied observation answers the stated B1 request, and how should the newer B2 revision be described?
Show the reference answer and checks
IS2 is the supplied observation matching Sand-A, measure V and basis B1. IS3's Sand-AX result belongs to a different grade. The newer Sand-A revision uses B2, so its value of 20 does not directly answer the B1 request without a documented equivalence or conversion. Explain the changed basis and seek clarification on which report the requester intends to use.
- Matching Sand-AX to Sand-A by a partial product-name string.
- Replacing the B1 observation with a newer B2 value while dropping the basis change.
Fields to preserve
- Requested grade and measure
- Defines the exact question rather than a name-only search.
- Test basis
- Makes differing scopes visible even when the units match.
- Revision relationship
- Shows that a later report exists and what changed.
- Conversion evidence
- Separates an authorized comparison from an assumed equivalence.
Test another case
Add a written request revision accepting B2 for Sand-A. Score whether the model now selects revision 2 for that revised request without rewriting the answer to the original B1 question.
Use artificial grades and values. Publish no real test thresholds, product recipes, customer requirements or proprietary laboratory results, and make no buyer acceptance decision.
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.