Abstract: A novel algorithm FFSPAN (fast frequent sequential pattern mining algorithm) is proposed to solve the problem that the computational complexity may become very high when mining long patterns in a sequence database. Traditionally, to judge whether a sub-sequence is frequent in a database, one need to compare the whole sub-sequence with every sequence in the original database, however the algorithm FFSPAN succeeds in solving the problem that in a sequence database, instead of searching a whole frequent sequence, we only need to search a frequent item or a frequent itemset. Moreover, the databases scanned via FFSPAN keep shrinking, which makes the algorithm more efficient when the sequential patterns are longer. Experiments on standard test data show that FFSPAN is very effective.

Key words: sequential pattern, long pattern, depthfirst, data mining