Using MapReduce framework for mining association rules

Abstract

Data mining in knowledge discovery helps people discover unknown patterns from the collected data. PIETM (Principle of Inclusion-Exclusion and Transaction Mapping) algorithmis a novel frequent item sets mining algorithm, which scans database twice. To cope with big transaction database in the cloud, this paper proposes a method that parallelizes PIETM by the MapReduce framework. The method has three modules. Module I counts the supports of frequent 1-item sets. Module II constructs transaction interval lists. Module III discovers all the frequent item sets iteratively. © 2013 Springer Science+Business Media Dordrecht.