An Enhanced Approach to Extract Top-K High Productive Itemsets

Abstract

The main purpose of this work is to develop a superior structure to extract top-K high productive itemsets. Here K is the picked portion of high productive itemsets that is to be established. High productive itemset tunneling is surely a prominent study in data mining but the factors for setting minimum utility margin are definitely a difficult task. In this work, an Enhanced approach to extract top-k high productive itemsets named Enhanced top-K high productive itemset tunneling (ETKU) is proposed. ETKU uses B+ Tree data structure instead of using a Utility Pattern Tree (UP-Tree) data structure that is used in existing Top-K high productive itemsets tunneling (TKU) method. Although TKU helps to reduce the time taken for the process of tunneling by reducing the total number of database scans to two, the complexity lies in the UP-Tree traversal for obtaining potential top-K high productive itemsets. B+ Tree used in ETKU does not have data associated with interior nodes so that more keys can fit into the memory. The leaf nodes of B+ Tree are linearly linked, so a full scan of a tree requires only one linear pass through all the leaf nodes.