A More General Form of Apriori and Its Application in Clustering Security Events

Abstract

Due to its excellent performance, Apriori is frequently adopted to discover frequent itemsets, from which strong association rules can be easily generated, from among massive amounts of transactional or relational data. In this paper, Apriori is reconsidered with a more abstract perspective of data space, and a more general form of the algorithm is proposed. As is shown in this paper, the newly proposed form of the algorithm can be effectively applied to more situations in which its primitive form does not work. The more general form of Apriori is successfully applied to the problem of clustering security events organized in a hierarchical manner, which illustrates its usefulness.