INDUSTRY FIELD GUIDE
Rough rice sample and milling-yield data for AI review
A rough-rice grower can show buyers how sample identity survives several measurements.
NAICS 111160 · Rice Farming
A rough-rice grower can show buyers how sample identity survives several measurements. USDA’s rice inspection handbook describes preparation and grading for multiple forms of rice. A training packet can test whether a model keeps a rough sample, a milled result, and any reported fractions tied to the same specimen. USDA rice inspection handbook
Use two mock sample tickets from the same delivery day. One specimen has a rough weight and a matching milled result. On the other, provide a whole-kernel figure whose sample number belongs to a different delivery. Write units, denominator, and measurement stage explicitly. Include a case where a component looks numerically plausible but cannot be used. A reviewer can score ID matching, denominator checks, and the ability to say when data is insufficient.
A lab or mill may control its grading worksheets even when a grower owns the rice. Use fictional tickets and owned photos first, and keep buyer deductions, cultivar contracts, and shipment identifiers out of scope.
Describe a rice inspection dataset
Original Fiund exercise
A plausible whole-kernel number is unusable when its sample belongs to another delivery
Original mock rice tickets use arbitrary mass units and a specified exercise denominator. They are not actual inspection sheets, grades or commercial deductions.
| Record | Source | What it records |
|---|---|---|
| RR1 | Sample A stages | Delivery D1, sample A: rough portion 100 mass units, matching milled output 70, matching whole-kernel component 55. |
| RR2 | Sample B packet | Delivery D1, sample B: rough portion 100, milled output 68. Attached whole-kernel card says sample B-old, delivery D0, component 54. |
| RR3 | Calculation definition | For this practice question, each output fraction uses its own sample's original rough portion. No B-to-B-old equivalence supplied. |
Question for the model
Which fractions can be computed for A and B, and can B's attached 54-unit component be used because it fits under the milled total?
Show the reference answer and checks
RR1 and RR3 support A's milled fraction of 70% and whole-kernel fraction of 55%, each over its 100-unit rough portion. RR2 supports B's milled fraction of 68%, but its attached 54-unit component belongs to another sample and delivery. Numerical plausibility does not resolve identity, so B's whole-kernel fraction remains unsupported. Preserve sample, stage and denominator before calculating rather than borrowing a neighboring result.
- Accepting a component because it is numerically plausible despite a delivery mismatch.
- Mixing stage-specific results or using a different sample's denominator.
Fields to preserve
- Delivery and sample identity
- Defines the specimen.
- Rough, milled and component stage
- Preserves the measurement relationship.
- Denominator rule and unmatched card
- Makes calculation limits explicit.
Test another case
Change the requested denominator to matching milled output. Require A's whole-kernel share to become 55/70, about 78.57%, with the new basis stated.
Use invented tickets and owned photos. Exclude buyer deductions, cultivar contracts and shipment IDs; mill and lab worksheet rights are separate from crop ownership.
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.