一种基于隐私保护的分布式聚类算法

Distributed Clustering Algorithm Based on Privacy Protection

针对水平划分的分布式数据库提出了一种基于隐私保护的分布式聚类算法PPDK-Means,该算法基于K-Means的思想实现分布式聚类,并且聚类过程中引入半可信第三方,应用安全多方技术保护本站点真实数据不被传送到其他站点,从而达到隐私保护的目的。理论分析和实验结果表明PPDK-Means算法是有效的。

This paper proposed algorithm PPDK-Means for privacy-preserving K-Means clustering over horizontal partitioned database. Using semi-trusted third party and secure multi-party technology, PPDK-Means does not transfer real data to other sites in clustering procedure. Theoretical analysis and experimental results show that algorithm PPDK- Means is effective and privacy preserving.