Fast approximate string matching with finite automata

We present a fast algorithm for finding approximate matches of a string in a finite-state automaton, given some metric of similarity. The algorithm can be adapted to use a variety of metrics for determining the distance between two words.