Fuzzy name matching for sanctions screening: normalisation, transliteration and Jaro-Winkler
Sanctions and PEP screening cannot rely on exact string matching, because one person may appear under many spellings, such as Muhammad Hussein, HUSSAIN, or Mohamed. The article proposes a three-layer approach: normalising names by removing case, accents, punctuation, titles and particles, mapping transliteration variants to a single form, and then scoring the remaining tokens with a similarity measure like Jaro-Winkler that ignores word order. It notes that a similarity threshold alone cannot distinguish true matches from namesakes, so date of birth and nationality are also needed. The author shares a tested Python implementation along with the trade-offs involved, including the risk of missing sanctioned individuals versus flagging innocent people.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.


Discussion (0)
Log in to join the discussion and vote.
Log in