INDUSTRY FIELD GUIDE
Fabric finishing and coating release records for AI evaluation
A test result that passed on one roll may belong to a different roll split from the same parent.
NAICS 313310 · Textile and Fabric Finishing Mills
NAICS 313320 · Fabric Coating Mills
A test result that passed on one roll may belong to a different roll split from the same parent. This is a data problem a finishing mill can demonstrate without sharing a bath recipe or a coating formulation. NIST’s manufacturing monitoring work explores linking process data with inspection and other evidence. It provides a research precedent for the method. A coated-fabric dataset would need specific buyer confirmation. NIST manufacturing monitoring
Build a hold case from two roll IDs
Make an invented traveler for a parent roll split into A and B. Give A a valid shade comparison or coating-test record under the active specification. Give B a result with a matching parent ID but an obsolete test revision. Mark which roll a reviewer would hold and which evidence would resolve the case. Attach a neutral coupon reference if an image supports the judgment. The model should trace lineage and explain why the apparent pass cannot be reused.
An approvable sample can omit customer color standards, restricted finish recipes, suppliers, and actual tolerances. If the buyer needs numerical ranges, agree first whether representative dummy values can serve the evaluation.
Describe the verification workflow
Original Fiund exercise
A coating report tied to one daughter roll
A fictional finishing ledger has two daughters after a roll split. The reviewed product is represented only by inert sample cards.
| Record | Source | What it records |
|---|---|---|
| F1 | Split ticket | Parent P17 yielded daughters D11 and D12. |
| F2 | Order revision log | D11: finish sheet rev B; D12: finish sheet rev C. |
| F3 | Shade sheet | Sample from D12 assessed against rev B. |
| F4 | Coating evidence | D12 coating check explicitly references rev C; no rev C shade check is recorded. |
Question for the model
What evidence is missing from the D12 packet?
Show the reference answer and checks
A shade check for D12 against rev C is missing. F3 ties the shade observation to the correct daughter but to the wrong reference for D12 under F2. F4 can populate the coating evidence slot only; neither shared parentage in F1 nor the coating check fills the shade slot.
- Accepting the latest-looking document without comparing the referenced revision
- Using D11's effective revision for both daughters
Fields to preserve
- daughter_roll
- Identifies the material under review after the split
- check_type
- Keeps coating and shade evidence separate
- reference_revision
- Tests applicability against the effective order document
Test another case
Move the revision transition inside D12 and provide a position map; compare only the marked portion to its effective reference.
Do not include production formulations, actual color references, customer orders or authority to release material.
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.