A fast updated algorithm to maintain the discovered high-utility itemsets for transaction modification. Issue 3 (August 2015)

• Record Type: Journal Article
• Authors: Lin, Jerry Chun-Wei; Gan, Wensheng; Hong, Tzung-Pei
• Is Part Of: Advanced engineering informatics. Volume 29:Issue 3(2015:Aug.)
• Page Start: 562
• Page End: 574
• Publication Date: 2015-08
• Subjects: Computer-aided engineering -- Periodicals; Engineering -- Data processing -- Periodicals; 620.00285
• DOI: 10.1016/j.aei.2015.05.003
• Languages: English
• ISSNs: 1474-0346

Abstract

<abstract xml:lang="en" abstract-type="author" id="ab005"> <title id="st005">Abstract</title> <sec> <p id="sp0005">High-utility itemsets mining (HUIM) is a critical issue which concerns not only the occurrence frequencies of itemsets in association-rule mining (ARM), but also the factors of quantity and profit in real-life applications. Many algorithms have been developed to efficiently mine high-utility itemsets (HUIs) from a static database. Discovered HUIs may become invalid or new HUIs may arise when transactions are inserted, deleted or modified. Existing approaches are required to re-process the updated database and re-mine HUIs each time, as previously discovered HUIs are not maintained. Previously, a pre-large concept was proposed to efficiently maintain and update the discovered information in ARM, which cannot be directly applied into HUIM. In this paper, a maintenance (PRE-HUI-MOD) algorithm with transaction modification based on a new pre-large strategy is presented to efficiently maintain and update the discovered HUIs. When the transactions are consequentially modified from the original database, the discovered information is divided into three parts with nine cases. A specific procedure is then performed to maintain and update the discovered information for each case. Based on the designed PRE-HUI-MOD algorithm, it is unnecessary to rescan original database until the accumulative total utility of the modified transactions achieves the designed safety bound, which can greatly reduce the computations of multiple database scans when compared to the batch-mode approaches.</p> </sec> </abstract>