Mining Multi-Relational Frequent Patterns in Data Streams
To the best of our knowledge, the problem of mining multi-relational frequent patterns in data streams is still unsolved up to now. To attack this problem, an algorithm RFPS, which is based on novel data synopsis and declarative bias, is proposed in this paper. By introducing a new data synopsis method, where period sampling is used, many samples checking operations are avoided. Meanwhile, lots of relation join operations are abridged by the utility of a new declarative bias, Join Tree, which makes the pattern refinement in RFPS more efficient. The theoretical analysis and experiments show that, the performance of RFPS is evidently better than static multi-relational frequent patterns mining algorithms, and the problem of mining multi-relational frequent patterns in data streams could be solved properly by this algorithm.
data mining multi-relational data streams frequent itemset period sampling
Wei Hou Bingru Yang Yonghong Xie Chensheng Wu
School of Information Engineering University of Science and Technology Beijing Beijing, China Beijing Municipal Institute of Science and Technology Information Beijing, China
国际会议
北京
英文
205-209
2009-07-24（万方平台首次上网日期，不代表论文的发表时间）