Industry field guide
Cinema composition and screening records for AI
Two digital cinema compositions can have the same public title and different audio or subtitle versions.
- 2022 NAICS
- 512131
- Practice pack
- Example and answer key (JSON)
- Updated
What this guide tests
Which composition matches the booking metadata, and does its inventory presence establish that the proposed playlist uses it?
See the specimen and fields · Work through the fictional records
Task and source context
Two digital cinema compositions can have the same public title and different audio or subtitle versions. Digital Cinema Initiatives documents composition and packaging concepts. A theater can test whether an assistant matches a practice booking to the intended composition without sharing a film or system access. Digital Cinema Initiatives specification
Invent two titles and two composition variants. Give one booking a language and accessibility requirement, plus a matching mock inventory record. Give another an asset with a similar name but the wrong track. Ask what the records support and what an authorized operator must confirm.
Do not include actual films, keys, credentials, screen schedules, supplier terms or customer data. Use independently written records and nonvalid identifiers. The model does not control playback.
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.
512131 · Motion Picture Theaters (except Drive-Ins)
This U.S. industry comprises establishments primarily engaged in operating motion picture theaters (except drive-ins) and/or exhibiting motion pictures or videos at film festivals, and so forth.
Build a pilot sample
Set up this pilot
Make dummy cinema composition IDs with two audio-track variants. Match the booking to the inventory, then add a playlist selecting the other version. Score availability and actual selection separately.
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
- Composition ID and track variants
- Disambiguates packages sharing a title.
- Booking requirements
- Defines the intended metadata match.
- Inventory versus playlist reference
- Separates availability from selection.
Use Make a segment-level rights inventory for media and publishing to turn the specimen into a brief with a defined task, source and review owner.
Fictional exercise
A matching public title can conceal the wrong track version
Nonvalid composition identifiers and invented booking records describe a fictional film. No media, keys, playback control or actual screen schedule is supplied.
| Record ID | Source | What it records |
|---|---|---|
| CI1 | Practice booking | Title Lantern; requested audio language L1 and caption variant C2. |
| CI2 | Dummy composition inventory | COMP-A: Lantern, L1/C1. COMP-B: Lantern, L1/C2. |
| CI3 | Playlist record | Proposed playlist references COMP-A. Inventory confirms COMP-B present, but no authorized playlist amendment recorded. |
Question for the model. Which composition matches the booking metadata, and does its inventory presence establish that the proposed playlist uses it?
Show the reference answer
CI1 and CI2 identify COMP-B as matching both L1 audio and C2 captions. COMP-A matches the title and language but has C1 captions. CI3's proposed playlist still names COMP-A; COMP-B's presence in inventory does not amend that reference. Flag the playlist-to-booking mismatch for an authorized operator rather than claiming the correct composition is selected or controlling playback.
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
- Matching on title and audio language while dropping the caption variant.
- Treating an available composition as the one referenced in the playlist.
Change the inputs and run it again
Add COMP-C with C2 captions but language L2. Require the model to check both dimensions instead of selecting whichever record contains C2.
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 independent records and dummy IDs. Exclude films, keys, credentials, supplier terms, real screening schedules and customer data.
Use original or explicitly cleared material and keep third-party music, performer data and unreleased content outside the package.
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
Troveo review: video licensing, business data and creator payouts
Compare the creator licensing pathway and the published steps between upload and payout.
Compare the approach
Protege review for AI teams and data owners
Compare its video or audio data-provider path and the questions to ask about permission evidence.
Related industry examples
Industry guide · NAICS 512132
Drive-in screening schedule revisions for AI testing
A changed outdoor showtime may appear in a notice while the earlier schedule remains in the booking tool.
Industry guide · NAICS 512120
Film distribution rights window data for AI review
An active exhibition window says something about distribution.
Industry guide · NAICS 512110
Film production shot and release records for AI
A production can own a camera file and still need separate approval for the people, material or locations shown.