Hits:

Indexed by:期刊论文

Date of Publication:2012-01-01

Journal:ICIC Express Letters, Part B: Applications

Included Journals:EI、Scopus

Volume:3

Issue:5

Page Number:1147-1155

ISSN No.:21852766

Abstract:This paper proposes a novel frequent itemsets mining algorithm based on transaction matrix, called TMA, to improve the efficiency of finding frequent itemsets for personalized recommendations. By constructing a data structure named Transaction Matrix, TMA creates candidate itemsets by scanning database only once. Meanwhile, a series of processes composed of AND and XOR operations is designed to generate candidate itemsets, which can compress candidate searching space and avoid redundant candidates by deleting infrequent itemsets during mining frequent itemsets. To evaluate the performance of our algorithm, experiments are carried out on four different datasets, and the results show that TMA outperforms three other algorithms in terms of the efficiency. Thus, TMA has a better applicability for precise personalized recommendation. ? 2012 ICIC International.