ACCURATE RETRIEVAL OF HIGH UTILITY ITEM SETS FROM THE TRANSACTIONAL DATABASE TO SUPPORT SERVICE ORIENTED COMPUTING

Abstract

In data mining, utility mining is a key field and high utility itemsets (HUIs) are revealed using this. From a multi-level dataset, HUIs are retrieved by this field. In a multi-level dataset, HUIs are retrieved using an existing algorithm called MUMA (multilevel utility mining algorithm). A tree structure is implemented in MUM algorithm and it is termed as MUMT (multilevel utility mining tree) for storing utility information of itemsets. For pruning the pattern, item profits are only considered in the existing work, which minimizes the accurate utility mining outcome and also ih has high computational overhead. For high utility mining, static data are considered in this work, so dynamic streaming data cannot be supported using this. Major objective of proposed research work is to develop a high utility mining extraction framework with reduced computation overhead. For the same, Dynamic Sliding Window Tree based Utility Mining Algorithm (DSWT-UMA) technique is introduced in this work. For high quality mining, considered the streaming data in this work. This work performs the construction of sliding tree by concerning time varying data for supporting high utility item mining form dynamic streamlining data. Based on profit and timing, considered the data pruning after constructing tree.This work prunes the item having old historic data and less profit. At last, based on this item counts, extracted the high utility items. In this work, JAVA is used for evaluating proposed associative classifier model. Better performance is exhibited by proposed technique than the existing work.