While much work has been done on determining if a document or a line of a document contains an exact or approximate match to a regular expression, less e ort has been expended in formulating and determining what to report as the match" once such a hit" is detected. For exact regular expression pattern matching, we give algorithms for nding a longest match, all symbols involved in some match, and nding optimal submatches to tagged parts of a pattern. For approximate regular expression matching, we develop notions of what constitutes a signi cant match, give algorithms for them, and also for nding a longest match and all symbols in a match.
Eugene W. Myers, Paulo Oliva, Katia S. Guimar&atil