Improvement and research on Aprioriall algorithm of sequential patterns mining

Computation of self-joining and pruning steps in the classical AprioriAll algorithm of sequence patterns mining is quite frequent while generating candidate sequence sets. Especially when the sequences are long, they will generate numerous of candidate sequence sets which leads to a combinatorial explosion phenomenon and makes the algorithm invalid. With the research on AprioriAll algorithm, self-joining and pruning strategies in AprioriAll have been modified, making candidate sets meet the dictionary sort feature after the operation of self-joining. Basing on generation rules of sub-sequence, frequent sequence sets have been cut before the operation of self-joining. The discuss result shows that the improved algorithm is much more efficient.