Fuzzy matching is an increasingly common and powerful technique used in computerized data analysis to find matches that may be less than 100% the same, but are close enough to warrant further scrutiny.
Fuzzy matching is most commonly performed on strings (names and addresses) but can also be performed on numeric values and dates to assess nearness rather than direct similarity.
Fuzzy matching is often used for fraud-based tests, tests of edit controls, and for comparing data stored between different systems (often new vs. legacy systems) as part of a data conversion.