Mining frequent sequential patterns under regular expressions: a highly adaptative strategy for pushing constraints∗

This paper introduces a new framework for the extraction of frequent sequences satisfying a given regular expression (RE) constraint. Contrary to previous work (SPIRIT algorithms), we represent REs by tree structures and our algorithm can choose dynamically an extraction method according to the local selectivity of the sub-REs. Interestingly, pruning can rely not only on the anti-monotonic minimal frequency constraint but also to the RE constraint that is generally not anti-monotonic. Preliminary experiments on synthetic data have shown that our algorithm takes the shape of the best algorithm from the SPIRIT family and even surpasses it.