一种车辆轨迹数据集的差分隐私发布技术  

A Differential Privacy Publishing Technology for Vehicle Trajectory Data Set

作　　者：凌虎 肖广兵[1] LING Hu;XIAO Guangbing(College of Automobile and Transportation Engineering,Nanjing Forestry University,Nanjing Jiangsu 210037)

机构地区：[1]南京林业大学汽车与交通工程学院,江苏南京210037

出　　处：《软件》2020年第11期153-156,204,共5页Software

摘　　要：本文试图将差分隐私模型应用于车辆时空轨迹数据集,提出一种分而治之的差分隐私处理方法。用评价平均密度的方法,将原车辆轨迹数据集分为路网与非路网车辆轨迹数据集,并且用时空聚合的方法进行车辆轨迹数据预处理。最后,分别用前缀树模型处理路网车辆轨迹数据集,用K-means 聚类模型处理非路网车辆轨迹数据集。从而,达到发布净化数据集的目的。This paper attempts to apply the diff erential privacy model to the vehicle spatiotemporal trajectory data set,and proposes a differential privacy processing method of divide and conquer.Using the method of evaluating average density,the original vehicle trajectory data set is divided into road network and non road network vehicle trajectory data set,and the vehicle trajectory data is preprocessed by the method of time-space aggregation.Finally,the prefi x tree model is used to process the road network vehicle trajectory data set,and the K-means clustering model is used to process the non-road network vehicle trajectory data set.Thus,the purpose of publishing the purifi cation data set can be achieved.

关 键 词：车辆轨迹数据 数据预处理 差分隐私保护 数据发布 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]