Effective algorithms for vertical mining probabilistic frequent patterns in uncertain mobile environments
Xiaomei Yu, Hong Wang, Xiangwei Zheng, Yilei Wang · International Journal of Ad Hoc and Ubiquitous Computing · 2016
Data uncertainty is inherent in mobile applications. The traditional methods of mining frequent patterns are confronted with enormous challenges in uncertain mobile environments. The present achievements have shown that vertical mining algorithms are promising in mining expected support-based frequent patterns from uncertain data, while they have not captured much attention in mining probabilistic frequent patterns. In this paper, we propose two vertical mining algorithms (UBEclat and NDUEclat) for probabilistic frequent patterns mining (FPM). The UBEclat algorithm is applied to discover exact probabilistic frequent patterns in uncertain environments, while the NDUEclat algorithm is suitable for mining probabilistic frequent patterns approximately in mobile environments with huge uncertain data. We test the two algorithms on real and synthetic datasets, and compare them with well-known FPM algorithms. The extensive evaluations show that the novel Eclat-based algorithms outperform the comparative ones in performance of efficiency and precision.