An Algorithm for Approximate Membership checking with Application to Password Security

Given a large set of words W, we want to be able to determine quickly whether a query word q is close to any word in the set. A new data structure is presented that allows such queries to be answered very quickly even for huge sets if the words are not too long and the query is quite close. The major application is in limiting password guessing by verifying, before a password is approved, that the password is not too close to a dictionary word. Other applications include spelling correction of bibliographic files and approximate matching.