Efficient Mining of High Utility Patterns over Data Streams with a Sliding Window Method

摘要: High utility pattern (HUP) mining over data streams has become a challenging research issue in mining. The existing sliding window-based HUP algorithms stream suffer from the level-wise candidate generation-and-test problem. Therefore, they need large amount of execution time and memory. Moreover, their structures are not suitable for interactive To solve these problems algorithms, this paper, we propose new tree structure, called HUS-tree (High Utility Stream tree) novel algorithm, HUPMS (HUP Mining data), streams. By capturing important information into an HUS-tree, our algorithm can mine all HUPs current window with growth approach. is very efficient Extensive performance analyses show that significantly outperforms algorithms.