An Algorithm of Fast Mining Frequent Itemsets Based on Sequence Number

Aiming to redundant candidate itemsets and repeated computing existing in presented mining algorithms, this paper proposes an algorithm of fast mining frequent itemsets based on sequence number. To fast execute double search, the algorithm adopts two methods of generating candidate itemsets, one is down search that generating subsets of non frequent itemsets, another is up search that computing their digital complementary sets. To only scan once database the algorithm uses character of attribute sequence number to compute support. The algorithm deletes repeated l-candidate itemsets generated by (l+1)-non frequent itemsets via locating subsets’ order, and also improves speed of generating candidate itemsets by computing subsets’ digital complementary set. The result of experiment indicates that the algorithm is faster and more efficient than presented algorithms of mining frequent itemsets.