Optimization of High Utility Itemset Mining from Large Transaction Databases on multi core processor

Abstract

High utility itemset mining is an emerging era that extends frequent itemset mining to identify itemsets in a transaction database with utility values associated with every item above a given threshold. Researchers recently proposed algorithm TWU (Transaction Weighted Utility) has anti-monotone property for pruning the datasets, but it is an overestimate of itemset utility that leads to more search space. In this paper we present an algorithm that takes features of CTU-PROL which is proposed by Researchers. It uses TWU with pattern growth based on a compact utility pattern tree data structure. Our algorithm runs on multi-core processor when the main memory is insufficient to deal with large datasets. An experimental result shows a remarkable speedup for large datasets than the previous algorithms. It can mine large data set more efficiently of both dense and sparse data.
DOI: 10.17762/ijritcc2321-8169.1506167