| Severity | Vendor | Invoice A | Invoice B | Amount | Why it was flagged |
|---|
Invoice numbers lose prefixes, punctuation, spaces and leading zeros, so INV-001023 and 1023 become the same. Vendor names drop case, punctuation and suffixes like Inc or LLC.
Invoices are grouped by amount, then compared within each group. Edit distance counts the character changes between two invoice numbers and turns it into a 0–100% score.
High: numbers match after cleaning. Medium: numbers are similar enough to be a typo. Low: worth a look, such as different numbers a few days apart, or sequential numbers a month apart that are probably recurring bills.