Original scientific paper

https://doi.org/10.17559/TV-20220205155952

Sequential Pattern Mining with Multidimensional Interval Items

Bob Chen ; School of Artificial Intelligence, Beijing Normal University, No.19, Xinjiekouwai St, Haidian District, Beijing, 100875
Weiming Peng ; School of Artificial Intelligence, Beijing Normal University, No.19, Xinjiekouwai St, Haidian District, Beijing, 100875
Jihua Song ; School of Artificial Intelligence, Beijing Normal University, No.19, Xinjiekouwai St, Haidian District, Beijing, 100875

Abstract

In real sequence pattern mining scenarios, the interval information between two item sets is very important. However, although existing algorithms can effectively mine frequent subsequence sets, the interval information is ignored. This paper aims to mine sequential patterns with multidimensional interval items in sequence databases. In order to address this problem, this paper defines and specifies the interval event problem in the sequential pattern mining task. Then, the interval event items framework is proposed to handle the multidimensional interval event items. Moreover, the MII-Prefixspan algorithm is introduced for the sequential pattern with multidimensional interval event items mining tasks. This algorithm adds the processing of interval event items in the mining process. We can get richer and more in line with actual needs information from mined sequence patterns through these methods. This scheme is applied to the actual website behaviour analysis task to obtain more valuable information for web optimization and provide more valuable sequence pattern information for practical problems. This work also opens a new pathway toward more efficient sequential pattern mining tasks.

Keywords

data mining; item intervals; prefixspan; sequential pattern mining

Hrčak ID:

279469

URI

https://hrcak.srce.hr/279469

Publication date:

17.6.2022.

Visits: 989 *