DMHUPS: Discovering Multiple High Utility Patterns Simultaneously

摘要

High utility pattern mining in transaction databases has emerged to overcome the limitation of frequent pattern mining where only frequency is taken as the measure of importance without considering the actual importance of items. Among existing state-of-the-art algorithms, some are efficient on sparse datasets and some are efficient on dense datasets. In this paper, we propose a novel algorithm called DMHUPS in conjunction with a data structure called IUData List to efficiently mine high utility patterns on both sparse and dense datasets. IUData List stores information of length-1 itemsets along with their positions in the transactions to efficiently obtain the initial projected database. In addition, DMHUPS algorithm simultaneously calculates utility and tighter extension upper-bound values for multiple promising candidates. Therefore, DMHUPS finds multiple high utility patterns simultaneously and prunes the search space efficiently. Experimental results on various sparse and dense datasets show that DMHUPS is more efficient than other state-of-the-art algorithms.