A novel approach for data stream maximal frequent itemsets mining

摘要

This paper proposes a novel algorithm AMMFI based on self-adjusting and orderly compound policy to solve the problems of existing algorithms for mining maximal frequent itemsets in a data stream. The proposed algorithm processes the data stream based on sliding window technique and scans data stream fragments single-pass to obtain and store frequent itemsets in frequent itemsets list. It then constructs a self-adjusting and orderly FP-tree, dynamically adjusts the tree structure with the insertion of itemsets, uses mixed subset pruning method to reduce the search space, and merges nodes with the same min_sup in identical branch. Finally, orderly compound FP-tree is generated and it avoids superset checking in the process of mining maximal frequent itemsets. Detailed simulation analysis demonstrates that the presented algorithm is of high efficiency of space and time and is more stable.