计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (2): 125-130.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

隐私保护轨迹数据发布的l-差异性算法

郭旭东,吴英杰,杨文进,王晓东   

  1. 福州大学 数学与计算机科学学院,福州 350108
  • 出版日期:2015-01-15 发布日期:2015-01-12

l-diversity algorithm for privacy preserving trajectory data publishing

GUO Xudong, WU Yingjie, YANG Wenjin, WANG Xiaodong   

  1. College of Mathematics and Computer Science, Fuzhou University, Fuzhou 350108, China
  • Online:2015-01-15 Published:2015-01-12

摘要: 针对基于传统的k-匿名模型下移动用户轨迹数据发布隐私保护算法有可能将相似度极高的轨迹匿名在同一个匿名集中从而导致可能出现的用户个人隐私泄露风险的不足。设计了一种新的轨迹数据发布隐私保护算法。该算法基于k-匿名模型,将轨迹所在的二维空间划分成大小相等的单元格,之后将由轨迹数据得到对应轨迹经过的单元格序列,从而定义轨迹k-匿名下的l-差异性,算法在满足k-匿名模型的前提下通过聚类的方法构建匿名集,并保证匿名集中的轨迹满足l-差异性标准,以达到降低由于差异性不足引起用户隐私泄露的风险的目的。实验结果表明,该算法是可行有效的。

关键词: 隐私保护, 差异性, k-匿名, 轨迹数据发布

Abstract: Based on k-anonymity model, the traditional algorithm which protects mobile objects’trajectory data when they are publishing has a possibility of leaking the objects’personal privacy. To solve this problem, this thesis designs a new kind of algorithm which can protect trajectory data privacy when publishing. This algorithm is based on k-anonymity, divides the two-dimensional space into cells of equal size, defines the standard of l-diversity under trajectory, structures anonymous set via clustering under the premise of k-anonymity model and makes sure that the trajectories which gather anonymously meet the standard of l-diversity so as to minimize the risk of leaking user’s privacy that caused by the lack of diversity. The experimental results show that this algorithm is feasible and effective.

Key words: privacy preservation, diversity, k-anonymity, trajectory data publication