Eliminating Redundant Association Rules in Multi-Level Datasets.
Gavin Shaw, Yue Xu, Shlomo Geva · QUT ePrints (Queensland University of Technology) · 2008
Association rule mining plays an important job in knowledge and information discovery and there are many approaches available. However, there are still shortcomings with the quality of the discovered rules. Often the number of the discovered rules is huge and many of them are redundant, especially in the case of multi-level datasets. Previous work has shown that the mining of non-redundant rules is a promising approach to solving this problem. However, work by Pasquier et. al. [14] and Xu & Li [17,18] is only focused on single level datasets. In this paper, we propose an extension to this previous work that allows them to remove hierarchically redundant rules from multi-level datasets. We also show that the resulting concise representation of non-redundant association rules is lossless since all association rules can be derived from the representation. Experiments show that our extension can effectively generate multilevel non-redundant rules.