Frequent Itemset Discovery in E-commerce Domain: A novel Approach

In this paper we propose a structure to find the frequent itemsets from the e-commerce data. Frequent itemset discovery is a heavily researched area in the data mining field. Due to the rapid growthof online business culture, there are plenty of data available in e-commerce web servers. The proposed structure explains how data is collected, cleaned and mined to find frequent itemsets from the ecommerce domain. A new and efficient ILLT (Indexed Limited Level Tree) algorithm developed for discovering the frequent itemsets discussed in this paper. This algorithm can be established for mininglarge item sets or for any n transaction and also it is applicable for online and offline process. ILLT algorithm works in two phases. First the transactional data is converted into three level compact tree structures. Then this tree is scanned to discover the frequent itemsets for the given support level. ILLT algorithm determines the frequent item sets in the given database without doing multiple scans and extensive computations. The computing result shows that the introduced algorithm is producing the same output as the existing Apriori algorithm, and it can avoid missed mining effectively.