摘要
隐私保护是当前数据挖掘领域中一个十分重要的研究方向。本文针对关联规则挖掘中如何保持隐私的问题,采用一种不依赖具体数据的随机正交变换方法,从而解决了在数据集容量很大的情况下,运算量大的问题,并使用传统隐私保护度评价方法与正交变换的方向隐私保护度相结合的方法评价变换的隐私保护度,进而使得结果更符合实际情况。理论分析和论证表明本文中的方法具有很好的隐私性、高效性和适用性。
Privacy preserving is an important direction for data mining research. This paper is concentrated on the issue of protesting the underly- ing attribute values when sharing data for association rules mining, adopts a random orthogonal transformation method without depending on any conerete data and thereby solve the computational problems when handling large data sets. And then evaluates the privacy preserving degree of the random orthogonal transformation using the combination of the traditional evaluation method for privacy preserving degree and the direction privacy preserving degree, thus makes the results be more in line with the actual situation. Theoretical analysis and demonstrations shows that the method in this paper has a very good privacy, efficient and applicability.
出处
《科技和产业》
2010年第1期75-79,共5页
Science Technology and Industry
基金
国家自然科学基金(70971059)
关键词
隐私保护
数据挖掘
关联规则
随机正交变换
privacy preserving
data mining
association rules
random orthogonal transformation
