An hourglass shop’s care sentences hide shop words such as pontil and anneal. A transformer token-tags those terms, and a bar of tag counts shows anneal almost never sits next to a still. How should that chart be read?
Select an answer to reveal the explanation.
Short Explanation
Highlighting rare shop words and counting them next to photos shows where the dictionary never met the camera. The bar is a coverage gap on the text arm—not a fusion order.
Full Explanation
When an hourglass shop’s care sentences hide terms such as pontil and anneal, and a transformer token-tags those terms so a bar of tag counts shows anneal almost never sits next to a still mention, that is a text-arm coverage gap revealed by turning tags into a simple visualization. Domain 5 uses NLP to expose missing co-occurrence—not to mandate fusion redesign.
Treating the bars as a Domain 2 mandate to early-fuse every tagged token fails because tag-count gaps are corpus coverage findings, not fusion-type orders. Proving ACE / Audio2Face must narrate every care sentence fails because avatar speech is unrelated to text-tag coverage next to stills. Deleting all stills that lack the word anneal fails by over-pruning the image arm instead of collecting or captioning the missing anneal–still pairs.
Exam caveat: token-tag bars diagnose text coverage and co-occurrence. Act with collection or caption repair; do not confuse gaps with fusion mandates or avatar requirements.
Operational check: export tag co-occurrence with still references; open tickets to add anneal-linked care photos or captions; re-bar after intake so the civic glass shop shows anneal appearing beside the visual arm where process docs claim it belongs.