INDUSTRY FIELD GUIDE
Tidal generation test-window data for AI
A test result may refer to a deployment window rather than the entire observation period.
NAICS 221118 · Other Electric Power Generation
A test result may refer to a deployment window rather than the entire observation period. DOE’s marine energy resources and repository document research datasets. A project can create an offline benchmark around correctly matching a result and an approved test period. DOE marine renewable energy repository
Invent two test windows and a report that supersedes the identifier on one. Include an observation that falls outside both windows. Ask which result belongs to each test and what source documents any revision. Grade the answer on record identity and uncertainty, with no model guidance on deployment.
Use cleared public research material where terms allow or wholly fabricated records. Keep actual device design, site details, network access, raw operation traces and project negotiations private.
Describe a research records exercise
Original Fiund exercise
Match fictional research records to a renamed test window
Original administrative test records define half-open time windows. They provide no device design, deployment plan, site information or operating traces.
| Record | Source | What it records |
|---|---|---|
| WINDOW-T1 | Approved practice register | Window A covers 10:00 through times before 11:00. Window B covers 11:00 through times before 12:00. |
| RENAME-T2 | Confirmed identifier correction | B is renamed C for the same period. A is unchanged. |
| OBS-T3 | Practice observation | An 11:30 observation retains the former B label. |
| OBS-T4 | Unassigned observation | A second observation is at 12:10, outside the listed periods. |
Question for the model
Which current test-window identifier corresponds to T3, and where can T4 be assigned?
Show the reference answer and checks
T3 falls in the period now named C. T2 supplies the documented B-to-C relationship, so the retained B label can be reported with its correction. T4 lies outside both approved windows and has no supported assignment in the packet. Do not extend C's period to absorb it. The rename does not change A or any stated time boundary.
- Treating a rename as a new or expanded deployment period.
- Assigning an outside observation to the nearest listed window.
Fields to preserve
- Window boundaries
- Keep start inclusion and end exclusion explicit.
- Identifier lineage
- Preserve the documented B-to-C rename.
- Observation time
- Check membership independently of a copied label.
Test another case
Add an observation at 11:00. The supplied half-open rule places it in C, not A.
Use fictional records or permitted public research material. Keep real designs, sites, networks, raw operations and commercial negotiations outside the example.
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.