Research on association rules based on Complex Networks

Abstract

In most cases, the association rules mining (ARM) will generate a large number of rules, most of which are of no interest. In order to find the interesting rules from the large-scale association rules more effectively, this manuscript presents a new method to research the association rules based on Complex Networks. The proposed method first illustrate the association rules by complex networks, and then puts forward a kind of community detection algorithm, which is aim to get the relationship and classification of the result rules. Thus we can locate the interesting patterns more quickly and accurately. Meanwhile, this method also provides a visualization process of the association rules. The experimental results demonstrate that this new method with complex networks can group the result rules and find out the interest patterns more intuitively and efficiently.