I would guess it could be used for use cases where you can't fail. For example recovering URLs from erroneous OCR scans, or unstructured text which might conta

I would guess it could be used for use cases where you can't fail. For example recovering URLs from erroneous OCR scans, or unstructured text which might contain errors. It could probably also be used immediately on user input to suggest possible corrections for invalid URLs.