摘要
数据挖掘中的隐私泄漏问题一直备受关注,在确保隐私的前提下达到最佳挖掘效果是近年来数据挖掘领域的研究热点之一。为防止在数据挖掘中发生隐私泄漏等问题,基于隐私保护框架,提出一种支持动态计算冲突度的高效的敏感规则清洗算法。在隐藏敏感规则的同时,动态调整冲突交易的冲突度,以尽量减少对非敏感规则误隐藏的可能性。理论分析与实验结果表明,给出的算法隐藏失败率为零,且大幅度降低了误隐藏率,有效保护了敏感规则,显著改善了算法的清洗效果。
The privacy leaking issue of data mining is always drawing tremendous attention. Realizing optimal mining effect without priva- cy leaking is one of the active issues in the field of data mining. In order to prevent privacy leaking during data mining, based on privacy enforcing framework, propose a sensitive rule sanitization algorithm that supports dynamic degree of conflict calculation, which reduces misses costs to the best via altering conflict transaction record dynamically while achieving good concealing purpose. Theoretical analysis and experiment results show that the presented algorithm can protect the sensitive rules effectively with no hiding failure and reduce re- markably the error hiding rate, which enhances the performance of the algorithm significantly.
出处
《计算机技术与发展》
2015年第2期126-130,共5页
Computer Technology and Development
基金
广西自然科学基金(2011GXNSFA018152)
教育部2013年国家级大学生创新创业训练计划(201310593028)
关键词
隐私保护
关联规则挖掘
动态冲突度
数据清洗
privacy preserving
association rule mining
dynamic degree of conflict
data sanitization
作者简介
田兴邦(1992-),男,CCF会员,研究方向为数据库信息安全;
华蓓,硕士,讲师,CCF会员,通信作者,研究方向为网络信息安全;
钟诚,博士,教授,博士生导师,CCF高级会员,研究方向为网络信息安全与并行分布计算。