Title: A novel approach for mining probabilistic frequent itemsets over uncertain data streams

Authors: Tianlai Li; Fangai Liu; Xinhua Wang

Addresses: School of Information Science and Engineering, Shandong Normal University, No. 88, East Wenhua Road, Lixia District, Jinan, 250014, China; Shandong Provincial Key Laboratory for Computer Network, No. 2008, Xinluo Road, Gaoxin District, Jinan, 250101, China ' School of Information Science and Engineering, Shandong Normal University, No. 88, East Wenhua Road, Lixia District, Jinan, 250014, China ' School of Information Science and Engineering, Shandong Normal University, No. 88, East Wenhua Road, Lixia District, Jinan, 250014, China

Abstract: With the growing popularity of internet of things (IoT) and pervasive computing, a large amount of uncertain data has been collected. Frequent itemsets mining has attracted much attention in database and data mining communities. Current methods exists some disadvantages, such as inaccurate, low efficiency, etc. To address this problem, we propose a novel approach, called uncertain pattern-slide window algorithm (UP-SW) is presented. In this algorithm, a new tree structure called USFP-tree is designed to save the redeveloped header table; the model of slide-window is adopted into the renewal process of mining result. The USFP-tree is structured based on dynamic array (ARRAY) and link information (LINK), as the slide-window slides, the mining result saved in USFP-tree is refreshed. The probabilistic frequent itemsets are obtained by traversing the final ARRAY of header table. Experimental results and theoretical analysis show that UP-SW has better performance than several other UP algorithms, especially on the mining efficiency and reducing the memory usage.

Keywords: data mining; uncertain data streams; probabilistic frequent item sets; sliding windows.

DOI: 10.1504/IJADS.2018.092794

International Journal of Applied Decision Sciences, 2018 Vol.11 No.3, pp.302 - 316

Available online: 05 Apr 2018 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article