A New Algorithm Based on Matrix for Mining Inter-Transaction Association Rules

Abstract

The traditional studies on mining association rules are on mining intra-transaction associations. In this study, we extend scope of mining association rules from traditional intra- transaction associations to inter-transaction associations. Mining inter-transaction associations poses more challenges on efficient processing than mining intra-transaction associations because the number of potential association rules becomes extremely large. In this study, we introduce the notion of inter- transaction association rule, define its measurements: support and confidence, then we design a new matrix data structure, called Co-Occurrence Matrix, in short COM, to store the data information instead of directly using the transactional database and develop an efficient algorithm MMIT(an acronym for Matrix Mining Inter-transaction) based on COM, for mining inter-transaction associations. We compare MMIT with FITI the best algorithms presented by other researchers in previous studies and demonstrate MMIT is more efficient than FITI.