Algorithm for mining maximal length frequent itemsets based on order FP-tree

The mining of frequent itemsets has been limited by the large number of resulting itemsets as well as the high computational cost. In many application domains, however, it is often sufficient to mine maximum length frequent itemsets. An order FP-tree-based algorithm is proposed for the mining problem. A field max-level is added in head-table to record the greatest height of item. In the mining process, only the item which max-level value is equal or greater than the length of existing maximum length frequent itemsets is traversed. Neither producing conditional pattern base nor constructing conditional frequent pattern tree recursively is needed, and the support of maximum length frequent itemsets is calculated. The experimental results show that the algorithm accelerates the speed to traverse the tree and improves the mining efficiency.