隐私保护的多数据源高投票率序列模式挖掘  

Mining High-voting Sequence Patterns with Privacy-preserving in Multi-databases

作　　者：张莹[1] 钟诚[1] 

机构地区：[1]广西大学计算机与电子信息学院,南宁530004

出　　处：《小型微型计算机系统》2015年第1期100-105,共6页Journal of Chinese Computer Systems

基　　金：广西自然科学基金项目(2011GXNSFA018152)资助;广西研究生教育创新计划项目(YCSZ2012007)资助

摘　　要：在多数据源环境中,通过计算数据库项集相似度,按照一定分类原则,设计了数据库数目和类个数可变的多数据源序列模式分类算法;在给出一种有效的多数据源高投票率序列模式挖掘方法的基础上,结合隐私保护要求,采取"分类—清洗—合成—挖掘"方法,提出一种隐藏敏感模式的多源高投票率序列模式挖掘算法。与不隐藏敏感模式的多源高投票率序列模式挖掘算法进行实验测试对比,结果表明:本文提出的算法只需花费额外少量的敏感模式匹配处理时间,可确保算法能够在挖掘得到全局高投票率序列模式的同时,隐藏敏感模式、保护多源数据中的隐私信息.In the multi-database environment, a multi-database sequential pattern classification algorithm with multiple databases and classes is designed by computing the similarity of item set in databases and according to a special classification principle;and an effi- cient mining high-voting sequential patterns method is given, according to the requirement of privacy preserving, a hiding sensitive pat- terns high-voting sequential patterns mining algorithm is presented by applying ＂classifying-cleaning-composing- mining＂ method. Compared with the multi-database sequential pattern mining algorithm with no-hiding sensitive patterns, the experimental results show that the proposed algorithm can not only mine the required global high-voting patterns but also hide the sensitive patterns and protect the privacy information in original multiple databases by spending extra very little time to process sensitive patterns matching.

关 键 词：多源数据 序列模式挖掘 高投票率 隐私保护 敏感模式隐藏 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]