A city-council minutes desk has five known speechwriters and a stack of unlabeled drafts. How should author attribution be implemented?
Select an answer to reveal the explanation.
Short Explanation
Five known speechwriters and a stack of unlabeled drafts: treat author attribution as multi-class text classification over those candidates. A chat guess, extractive QA, an NCCL topology, or a video deepfake detector is not that head.
Full Explanation
Author attribution is official as a classifier over a known author set: one class per speechwriter, trained or served as text classification. Asking a chat model who probably wrote it, without a head, is not the named task. Extractive QA points at a span, NCCL topology is training plumbing, and a video deepfake detector is the wrong modality. Build the multi-class head, then label the unlabeled drafts.