Industry field guide
Heavy civil earthwork lift records for AI review
A test can refer to one layer of placed material while the report lists another.
- Sector
- Construction
- 2022 NAICS
- 237990
- Practice pack
- Example and answer key (JSON)
- Updated
What this guide tests
Which exact observation has a documented completed recheck, and does the shared point label close the L2 review?
See the specimen and fields · Work through the fictional records
Task and source context
A test can refer to one layer of placed material while the report lists another. USACE publishes levee construction and construction-control materials. A civil contractor can design a fictional quality record exercise where the model must link each observation to the correct layer and review. USACE levee construction manual
Use an invented cross-section with two training layers. Give one an initial test, documented rework and a matching recheck. Add a separate test with a similar identifier but a different layer. Ask which source supports each status and what the file cannot show. Have the reviewer score the association and chronology independently.
Keep actual flood infrastructure details, material properties, coordinates, equipment settings, and restricted engineering documents private. The exercise does not establish structural safety.
Industry research
The U.S. Census Bureau’s industry definitions establish which activities these codes cover. The workflow sources in this guide describe the task or record type. Fiund’s pilot below uses fictional material to test a specific decision within that scope.
237990 · Other Heavy and Civil Engineering Construction
This industry comprises establishments primarily engaged in heavy and civil engineering construction projects (excluding highway, street, bridge, and distribution line construction).
Build a pilot sample
Set up this pilot
Draw two fictional earthwork layers with the same point label. Record rework and independent recheck for only one layer and grade the composite identifier.
Prepare the test packets
- Start with the worked case. Use the fictional record types below as a guide to the source files you need. Give each packet a stable ID and save its source versions with it.
- Change one fact. Make one packet with enough evidence to answer the question and one where the evidence points to a different answer. Keep the other fields fixed so a reviewer can tell what changed.
- Remove the decisive evidence. Give the reviewer a third packet with a plausible near-match in its place. Have them write what can still be established and what evidence would resolve the rest.
Fields the reviewer needs
- Layer-plus-point identity
- Disambiguates repeated location labels.
- Initial, rework and recheck events
- Preserves the chronology supporting a status.
- Review outcome source
- Keeps an administrative summary from replacing the underlying test.
Use How construction and trade teams can contribute data for AI to turn the specimen into a brief with a defined task, source and review owner.
Fictional exercise
A recheck closes only the layer and observation it names
An abstract cross-section contains training layers L1 and L2. Invented test codes and dispositions measure association and event sequence without evaluating real earthwork.
| Record ID | Source | What it records |
|---|---|---|
| EL1 | Initial practice test | Layer L1, point Q: status needs review under test T1. |
| EL2 | Rework and recheck record | Layer L1, point Q: rework logged, then reviewed recheck T2 records exercise status complete. |
| EL3 | Separate test | Layer L2, point Q: test T2-B remains needs review. A summary shortens both point labels to Q. |
Question for the model. Which exact observation has a documented completed recheck, and does the shared point label close the L2 review?
Show the reference answer
EL1 and EL2 establish a sequence for L1 at point Q: initial review flag, logged rework and completed reviewed recheck T2. EL3 is L2 at point Q with its own open status. The shortened Q label does not erase the layer distinction. Report the completed L1 sequence and preserve L2's open review rather than applying one layer's outcome to both.
How to check the result
Ask for the answer, the record IDs that support it and any fact the evidence leaves unresolved. Have a reviewer compare the response against the reference before deciding whether the sample is ready to expand.
Errors to flag in this case
- Joining tests on point Q alone while dropping the layer key.
- Treating a rework log as the completed recheck before the reviewed observation occurs.
Change the inputs and run it again
Make the recheck timestamp precede the recorded rework event. Require an explicit chronology flag rather than a silently reordered completion narrative.
Score the decision and the evidence separately
Save the reviewer’s expected answer, supporting record IDs and unresolved fields in a separate answer key. Score the model on the decision, its citations and its handling of the deliberate gap. A correct answer citing a different case counts as an evidence error.
Before expanding collection, have a second reviewer run the same checks on the held-out packet. Keep related versions, views and exports of the same case together so they stay on one side of the training and evaluation split.
Check permission and scope
Use synthetic cross-sections and status labels. Exclude real infrastructure geometry, material properties, coordinates, equipment settings and engineering documents. No structural-safety conclusion is requested.
Use fictional coordinates or cleared plans. Avoid active infrastructure maps, routing details and site security information.
Describe an authorized archive through Fiund’s existing-data assessment, or describe a new capture capability through the collection brief. The contributor agreement covers the terms for participation.
Sources and further reading
Resources for this workflow
Company and program reviews
Use these reviews to compare sourcing, capture or evaluation approaches. Confirm the applicable program, accepted sources and terms with the provider.
Compare the approach
micro1 review for experts, trade footage and company data
Compare how a trade recording project differs from an expert role or a company data partnership.
Compare the approach
Scale AI review: data partnerships, Outlier and buyer services
Read the documented operational-data route for established field-service businesses.
Related industry examples
Industry guide · NAICS 237110
Water and sewer installation evidence for AI review
A photo taken before cover can document a connection without identifying which segment it belongs to.
Industry guide · NAICS 236115
Custom home selection and installation records for AI
An owner can approve a window selection while the site packet still shows the prior opening.
Industry guide · NAICS 237130
Power and communications conduit identity data for AI
Two conduit labels may appear interchangeable until the reviewer checks the drawing revision and the observed ID.