¹College of Information Engineering, Ningbo Dahongying University, Ningbo, Zhejiang, China 315175
²School of Computer Science and Technology, Faculty of Electronic Information and Electrical Engineering, Dalian University of Technology, Dalian, Liaoning, China 116024
³School of Innovation and Experiment, Dalian University of Technology, Liaoning, China 116024

---

Abstract—In this paper, we study the problem of finding frequent itemsets from uncertain data streams. To the best of our knowledge, the existing algorithms cannot compress transaction itemsets to a tree as compact as the classical FPTree, thus they need much time and memory space to process the tree. To address this issue, we propose an algorithm UDS-FIM and a tree structure UDS-Tree. Firstly, UDS-FIM maintains probability values of each transactions to an array; secondly, compresses each transaction to a UDS-Tree in the same manner as an FP-Tree (so it is as compact as an FP-Tree) and maintains index of probability values of each transaction in the array to the corresponding tail-nodes; lastly, it mines frequent itemsets from the UDSTree without additional scan of transactions. The experimental results show that UDS-FIM has achieved a good performance under different experimental conditions in terms of runtime and memory consumption.

Index Terms—frequent itemset, frequent pattern, uncertain dataset, data streams, data mining

[PDF]

Cite: Le Wang, Lin Feng, and Mingfei Wu, "UDS-FIM: An Efficient Algorithm of Frequent Itemsets Mining over Uncertain Transaction Data Streams," Journal of Software vol. 9, no. 1, pp. 44-56, 2014.