An Algorithm of Mining Long Frequent Itemsets Based on Digit Sequence

In order to fast generate candidate frequent itemsets, avoid redundant calculation and reduce the time of scanning database, this paper proposes an algorithm of mining frequent itemsets based on digit sequence, which is suitable for mining long frequent itemsets. The algorithm firstly turns all transactions into digital transactions by binary, and then computing digit sequence of every attribute item. Finally, the algorithm uses the method of forming digital pure subset of digital transaction to generate candidate frequent itemsets by declining the value of digital pure subset, and uses the method of computing dimension of digit sequence of attribute items to compute support of candidate frequent itemsets, this method is used to reduce the time of scanning transactions database. The algorithm only scans once database when mining all these association rules, which is different from presented algorithms of mining long frequent itemsets. The experiment indicates that the efficiency of the algorithm is faster and more efficient than presented algorithms of association rules mining.