Structured Data Needs Semantic Validation

Imagine OCR reads:
31.50
instead of:
37.50
The generated CSV can still be perfectly valid.
The parser won't necessarily know that the number is wrong.
That's why OCR validation should focus on high-impact fields:
numbers
dates
coordinates
IDs
totals
currency
field mappings
LoveOCR's documentation explicitly recommends reviewing these fields before important workflows.