Logical Itemset Mining Implementation on Hadoop

Frequent Itemset Mining (FISM) finds the large and frequently occurring items from the datasets using Apri-ori algorithm. The FISM framework does not addresses two major properties that are Mixture-of property(more than one customer intent) and Projection-of property. To overcome the problems of irrelevant and non ac-tionable data and also to address the properties men-tioned above, Logical Itemset Mining (LISM) frame-work is introduced. LISM finds logical itemsets from the data which helps in eliminating non actionable data but at the same time keeps data which is logically connected. LISM not only finds logically con-nected items but aso items which are rarely occurring but logically connected are also discovered. LISM also addresses the Mixture of property and Projection of property which are not very well addressed in FISM. General Terms Market Basket, Mixture of property, Projection of property