A new efficient approach for mining uncertain frequent patterns using minimum data structure without false positives

doi:10.1016/j.future.2016.09.007

2017-03-01

PRE

Unil Yun *

DOI：10.1016/j.future.2016.09.007

原文链接

原文求助

摘要

En 中文

The concept of uncertain pattern mining was recently proposed to fulfill the demand for processing databases with uncertain data, and various relevant methods have been devised. However, previous approaches have the following limitations. State-of-the-art methods based on tree structure can cause fatal problems in terms of runtime and memory usage according to the characteristics of uncertain databases and threshold settings because their own tree data structures can become excessively large and complicated in their mining processes. Various approximation approaches have been suggested in order to overcome such problems; however, they are methods that increase their own mining performance at the cost of accuracy of the mining results. In order to solve the problems, we propose an exact, efficient algorithm for mining uncertain frequent patterns based on novel data structures and mining techniques, which can also guarantee the correctness of the mining results without any false positives. The newly proposed list-based data structures and pruning techniques allow a complete set of uncertain frequent patterns to be mined more efficiently without pattern losses. We also demonstrate that the proposed algorithm outperforms previous state-of-the art approaches in both theoretical and empirical aspects. Especially, we provide analytical results of performance evaluation for various types of datasets to show efficiency of runtime, memory usage, and scalability in our method. (C) 2016 Elsevier B.V. All rights reserved.

Keyword:

Correctness

Data mining

Existential probability

Frequent pattern mining

Uncertain pattern