A copperplate shop needs a label that is true only of the pair—this motto belongs on this plate—not of either file alone. Where does that annotation belong?
Select an answer to reveal the explanation.
Short Explanation
“This motto fits this plate” is about the pair, not about text or image alone. That joint label is multimodal data work sitting on the paired example.
Full Explanation
A label that is true only of the pair—this motto belongs on this plate—is joint cross-modal labeling on the paired multimodal example. Placing that annotation on the paired example is correct because neither file alone carries the belonging relation. Civic provenance labels (this caption authenticates this still; this note belongs on this asset tag) must live on the joined example so auditors can see pair-level truth, not orphaned text claims.
Putting the label on the motto text arm only, ignoring the plate still, fails because the truth is relational across modalities. Parking it only on a Domain 5 accuracy chart after training fails by delaying pair-level ground truth until evaluation. Treating it only as a Domain 2 definition of early versus late fusion fails by confusing labeling location with architecture vocabulary.
Exam caveat: pair-only truths are multimodal annotations, not single-arm fields. Operational check: store civic joint labels on the paired example record, require both arms present, and review belonging claims against both files. Keep municipal reviewers in the loop whenever modality claims, fusion labels, or agent scripts are promoted from lab notes into the production corpus used by public-facing services.