面向数据直方图发布的差分隐私保护综述

随着大数据时代的到来，信息安全也日益成为了人们关注的话题和重点。与匿名隐私保护相比，差分隐私保护作为一种新的隐私保护技术，能抵抗假设攻击和背景知识攻击。差分隐私保护的直方图发布能够直观表示数据的发布信息，针对国内外在静态数据集和动态数据流方向上的数据直方图发布的差分隐私保护研究现状进行介绍，讨论有关静态数据集下直方图存在长区间添加噪声而导致的噪声累积、数据可用性低，以及动态数据流下隐私预算容易耗尽问题的解决方法，对基于直方图的差分隐私保护各相关算法进行对比与分析，最后总结出目前差分隐私保护技术的应用及未来的研究方向。

Survey of Differential Privacy for Data Histogram Publication

With the advent of the era of big data, information security has become the topic and focus attention. Compared with the anonymous privacy protection, differential privacy protection as a new privacy protection technology can resist assume attacks and background knowledge attacks. Histogram publication under differential privacy protection can give a visual representation for data release. This paper will introduce the technologies of differential privacy protection for data histogram publication in the static and dynamic data sets in the domestic and foreign. Then it will discuss the solutions to the noise accumulation and low data availability resulted by adding noise in long range in histogram in static data sets, and privacy budget being easy to run out in dynamic data sets. Some comparisons and analysis among main differential privacy protection technologies for histogram publication are made. Finally the paper summarizes some of the current applications and potential future research directions for differential privacy protection technology.