An Efficient Algorithm for Hierarchical Online Mining of Association Rules

Several multi-pass algorithms have been proposed for Association Rule Mining from static repositories. However, such algorithms are incapable of online processing of transaction streams. In this paper we introduce an efficient single-pass algorithm for mining association rules, given a hierarchical classification amongst items. Processing efficiency is achieved by utilizing two optimizations, hierarchy aware counting and transaction reduction, which become possible in the context of hierarchical classification. We also propose a modified algorithm for the rule generation phase which avoids the construction of an explicit adjacency lattice.