BIDEL algorithm based on the ezpansion of position
Pattern mining on frequent closed sequence is one of hot spots in data mining. Aimed at the set of frequent closed sequential patterns being mined more efficient, PBIDE is proposed, which is the BIDE algorithm based on the expansion of position. The position information, expansion event, and optimization strategy are defined. Firstly, the numbers of different sequences identities are calculated in the position information of each event. And frequent 1-sequence can be obtained by comparing the threshold with their numbers. Secondly, according to the idea of prefix and adjacent position expanding, for every frequent 1-sequence, the expansions of position are checked directly. The expansions of position include three directions: backward, forward and medially. That the pseudo-projection database is scanned repeatedly is avoided. The optimization strategy is applied to reduce the time and memory consumed in the process of pre-expansion and expansion. So, the set of frequent closed sequential patterns will be mind more efficient than before. The experimental results show that PBIDE has significantly lower running time costs than BIDE, especially on the intensive datasets.
pattern mining closed sequence position ezpanding
Haitao He Jia Zhang Jiadong Ren
College of Information Science and Engineering Yanshan University Qinhuangdao, China
国际会议
Second International Symposium on Electronic Commerce and Security(第二届电子商务与安全国际研究大会)(ISECS 2009)
南昌
英文
820-824
2009-05-22（万方平台首次上网日期，不代表论文的发表时间）