基于时间衰减模型的模糊会话关联规则挖掘算法

发布时间：2018-11-03 07:13

【摘要】：现有的关联规则挖掘算法没有考虑数据流中会话的非均匀分布特性和历史数据的作用,并且忽略了连续属性处理时的尖锐边界问题。针对这些问题,提出一种基于时间衰减模型的模糊会话关联规则挖掘算法。针对数据流中会话的非均匀分布特性,基于时间片对会话进行划分,完整地保留了时间片内会话之间的相关性信息,采用模糊集对会话的连续属性进行处理,增加了规则的兴趣度和可理解性。在考虑历史数据作用和允许误差情况的基础上,基于时间衰减模型挖掘数据流中的临界频繁项集和模糊关联规则。实验结果表明,该方法在提高时间效率、降低冗余率和增加规则兴趣度方面存在明显优势。
[Abstract]:The existing association rules mining algorithms do not consider the non-uniform distribution of sessions and the role of historical data in the data flow, and ignore the sharp boundary problem in continuous attribute processing. To solve these problems, a fuzzy session association rule mining algorithm based on time attenuation model is proposed. In view of the non-uniform distribution of sessions in the data stream, the sessions are partitioned based on the time slice, and the correlation information between the sessions within the time slice is kept completely. The fuzzy set is used to deal with the continuous attributes of the session. It increases the interest and comprehensibility of the rules. The critical frequent itemsets and fuzzy association rules in the data stream are mined based on the time attenuation model on the basis of the historical data function and the allowable errors. Experimental results show that this method has obvious advantages in improving time efficiency, reducing redundancy and increasing rule interest.
【作者单位】：解放军信息工程大学;河南省信息安全重点实验室;
【基金】：国家“863”计划资助项目(2012AA012704) 国家“973”计划资助项目(2011CB311801) 郑州市科技领军人才项目(131PLJRC644)
【分类号】：TP311.13

【相似文献】