INDUSTRY FIELD GUIDE
Construction sand and gravel fraction records for AI
A result for a separated fraction can appear beside a record for a combined sample.
NAICS 212321 · Construction Sand and Gravel Mining
A result for a separated fraction can appear beside a record for a combined sample. USGS publishes construction sand and gravel data. A producer can create a fictional specimen and fraction inventory to test whether a model assigns each result to the right portion. USGS construction sand and gravel information
Assign two practice portions to one synthetic sample. Give one a reviewed measurement and the other only a combined value from the parent. Ask which figures can be attributed to each fraction and which field remains unknown. Add an incompatible unit to test that the assistant stops before combining values.
Keep actual material samples, quality records, buyer specifications, pricing, plant information and site locations out of the packet. The exercise does not approve a material for a project.
Describe a fraction-trace exercise
Original Fiund exercise
A parent aggregate does not label every separated fraction
This invented sand-and-gravel packet uses two practice fractions and a combined parent record. The quantities are arbitrary and provide no evidence that a material meets a construction specification.
| Record | Source | What it records |
|---|---|---|
| SG1 | Practice separation ledger | Parent S-10 contains fractions S-10-A and S-10-B. |
| SG2 | Reviewed fictional result | S-10-A has score 7 units under test T; S-10-B has no individual score. |
| SG3 | Combined parent summary | S-10 reports 12 units for the combined sample. A separate entry lists 5 percent without a fraction identity. |
Question for the model
What score is directly supported for each fraction, and can the parent total or unassigned percent fill B's missing result?
Show the reference answer and checks
SG2 directly supports a test-T score of 7 units for S-10-A. No individual score for S-10-B is supplied. SG3's combined value is not an attribution rule, and the packet does not state that scores are additive. Subtracting 7 from 12 would invent B's result. The unassigned 5 percent has neither compatible units nor a verified fraction link.
- Deriving B=5 from a parent score without an additivity rule.
- Treating 5 percent as 5 test units and assigning it to B.
Fields to preserve
- Parent and fraction IDs
- Preserves the separation relationship.
- Individual versus combined scope
- Distinguishes a measured fraction from a parent summary.
- Measure definition and unit
- Identifies whether arithmetic is meaningful.
- Reviewer assignment
- Records which result was actually checked for which fraction.
Test another case
Provide an explicitly additive mass inventory with matching units and all but one component. Compare whether the model can calculate a remainder there while declining to infer a nonadditive test score here.
Keep real quality records, plant data, customer specifications, pricing and locations out of the exercise. No acceptance or engineering conclusion is requested.
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.