Glossary
Fuzzy Matching (Sanctions Screening)
Name-matching that tolerates spelling, transliteration, and formatting differences to catch sanctioned parties.
Definition
Fuzzy matching is the use of approximate string-matching algorithms to compare a counterparty name against a sanctions list while tolerating differences in spelling, transliteration, formatting, name order, and missing name components.
Regulatory context
OFAC expects screening to catch matches beyond exact spelling. Names on the SDN list frequently include aliases, transliterations from non-Latin scripts, and 'also known as' entries. Effective screening requires matching across this full surface area.
Who it affects
Anyone screening names (as opposed to wallet hashes). Wallet matching is exact; name matching is inherently approximate and requires a fuzzy layer plus human review of matches.
Relevance to AI agents
Autonomous agents screening names must apply fuzzy matching and route any non-trivial match to a human reviewer before payment. Agents must not auto-pay on 'no exact match' — that defeats the purpose of name screening.
SanctionsAI coverage
SanctionsAI applies fuzzy matching over the SDN list including aliases and transliterations. The response flags the matched name and confidence so a human can confirm or clear.
FAQ
1. Why is fuzzy matching needed for sanctions screening?
Because sanctioned parties use aliases, transliterations, and variant spellings. Exact matching alone misses most real-world matches.
2. What is a false positive in fuzzy matching?
A match that is mathematically similar but refers to a different person. False positives are why human review of name matches is standard practice.
Screen a counterparty free
OFAC sanctions screening in under 100ms. Free tier: 5 checks/day.
Screen a wallet free → See pricing