OCR files
ALTO word-confidence review
List words below a chosen WC threshold and keep missing confidence separate.
Your inputs
Start with an exampleYour result
Example resultAa — A small example of your colors.
How it works
Read each String’s WC attribute, where 0 means unsure and 1 means sure. Preserve IDs, text and any stated word box. Missing confidence is unknown.
With threshold 0.8, the sample selects WC 0.5, retains WC 0.9 outside the list and reports one unknown WC.
Limits
No OCR is run and no recognition-accuracy percentage is inferred. WC scores can differ between engines. Character confidence CC has different semantics and is not treated as WC.