Flexible Musi Retrieval in Sublinear Time

Abstra t. Musi sequen es an be treated as texts in order to perform musi retrieval tasks on them. However, the text sear h problems that result from this modeling are unique to musi retrieval. Up to date, several approa hes derived from lassi al string mat hing have been proposed to ope with the new sear h problems, yet ea h problem had its own algorithms. In this paper we show that a te hnique re ently developed for multipattern approximate string mat hing is exible enough to be su essfully extended to solve many di erent musi retrieval problems, as well as ombinations thereof not addressed before. We show that the resulting algorithms are lose to optimal and mu h better than existing approa hes in many pra ti al ases.