【作者】 严澄； 胡天磊； 陈珂； 陈刚；

【Author】 Yan Cheng,Hu Tianlei,Chen Ke,and Chen Gang (College of Computer Science and Technology,Zhejiang University,Hangzhou 310027)

【机构】 浙江大学计算机科学与技术学院；

【摘要】 数据流中的关联规则在预测和在线分析系统中有重要应用.现有的研究大多集中在事务数据模型上,鲜有对数据项之间的关联规则挖掘.由于数据的实时性特点,用户又往往对新产生的数据所包含的信息更感兴趣.为了实时而准确地挖掘最近一段时间内数据项间的关联规则,提出了MARSW(mining association rules on sliding window)算法,利用滑动窗口模型对数据流进行关联规则挖掘.MARSW算法在给定的误差范围内,能够有效去除历史数据的影响,并以有限的空间代价快速挖掘大量数据间存在的关联规则.大量仿真实验结果表明,MARSW算法具有较高的效率和优良的可扩展性.更多还原

【Abstract】 Association rules in data stream places an important role in prediction and online analysis systems.Most existing researches are focusing on transaction data model;few are mining the association rules between elements occurring in data stream.Due to the characteristics of real-time data,people are more interested in the information of recent data than that of the old.In order to mine association between elements,an algorithm is proposed to report the association rules under the sliding window model.The algorithm can effectively eliminate the impact of historical data and quickly mine association rules between large amounts of data at limited space cost within the given error band. Experimental results show that the proposed method is efficient and scalable.更多还原