Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/38065
標題: MINING HYBRID SEQUENTIAL PATTERNS BY HIERARCHICAL MINING TECHNIQUE
作者: Kuen-Fang Jea 
賈坤芳 
Lin, K.C.
Liao, I.E. 
廖宜恩 
關鍵字: Hybrid sequential pattern;Mining;Hierarchical mining technique;efficient algorithm;discovery;rules
Project: International Journal of Innovative Computing Information and Control
期刊/報告no:: International Journal of Innovative Computing Information and Control, Volume 5, Issue 8, Page(s) 2351-2367.
摘要: 
Unlike sequential patterns, hybrid sequential patterns display not only the path but also the relationship among transaction items. The information provided by the collection of hybrid sequential patterns is useful in improving the analysis of marketing strategies, such as, browsing web pages, discovering customers' behaviors and so on. The process of mining hybrid sequential patterns in a database, however, becomes Complicated by the huge number of candidate patterns. In this paper, we propose a hierarchical mining technique to deal with this complexity. The unique features of this new technique include: counting hybrid sequential patterns by class, and examining database transactions in a top-down manner. This results in scanning a database, at most, twice. Using the technique, we develop an efficient mining algorithm, and conduct a simulation to study its performance. There are three major contributions in this paper. First, our proposed pattern-class concept provides a new way to count a group of patterns simultaneously. Second, we propose a novel decomposition model to lower the I/O cost in counting patterns from a large database. And third, we prove the correctness of counting patterns in the pattern decomposition model in this paper.
URI: http://hdl.handle.net/11455/38065
ISSN: 1349-4198
Appears in Collections:資訊科學與工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.