摘要：

km-匿名可以抵制长度为m的背景知识攻击,然而现有的匿名化算法在泛化处理时,优先选择支持度最小的位置点进行处理,未考虑泛化造成的变形度。随着m值的增大,轨迹变形度会变大。针对该问题,提出2种匿名化算法:最小变形度贪心算法和基于先验原则的最小变形度贪心算法,2种算法优先选择变形度最小的位置点进行泛化,使得泛化所造成的变形度更小,并给出匿名轨迹可用性度量方法,对数据可用性和算法效率进行分析。实验结果表明,与现有的匿名化算法相比,2种算法均可生成可用性更高的匿名轨迹。

关键词: 隐私保护, km-匿名, 轨迹, 背景知识攻击, 点泛化变形度

Abstract:

km-anonymity can resist background knowledge attacks of m-length subjectory.However,the existing anonymized algorithms select the minimum support point,not the minimum distortion point to generalize.Therefore,with increasing of m,the distortion of the trajectory tends to be larger.To address the problem,this paper proposes two kinds of anonymized algorithms:one is the minimum distortion greedy anonymized algorithm,the other is the minimum distortion greedy anonymized algorithm with apriori principles.These two algorithms both take consideration of the effect of the generalizing distortion,and choose the minimum distortion point to generalize,which can cause less distortion.It also proposes a method to measure the anonymous trajectory utility.It analyses the data availability and algorithm efficiency.Experimental results show that the proposed algorithms can generate anonymous trajectory with higher trajectory utility than existing anonymized algorithms.

Key words: privacy preservation, km-anonymity, trajectory, background knowledge attack, point generalized distortion degree

中图分类号: 

• TP301.6