Differential Privacy for Social Science Inference
Differential Privacy for Social Science Inference
复制标题
社会科学推理的差异隐私
DOI：
10.2139/ssrn.2676160
复制
发表时间：
2015
期刊：
影响因子：
--
通讯作者：
Gary King
中科院分区：
文献类型：
--
作者：
Vito D'Orazio;James Honaker;Gary King
社会科学家通常希望分析包含必须保密的敏感个人信息的数据。然而，试图保护隐私的常见数据共享技术要么带来巨大的隐私风险，要么带来巨大的信息损失。大量文献表明，数据发布的匿名技术通常容易受到重新识别攻击。汇总信息可以减少但不能防止这种风险，同时也会降低数据对研究人员的效用。即使在不公开数据的情况下发布统计估计，也不能保证敏感的个人信息不会被泄露。差分隐私源于密码学，是隐私保护的一种正式的数学概念。它带来了可证明的保证，即任何报告的结果都不会泄露任何一个人的信息。在本文中，我们详细介绍了安全策展人界面的构建，通过该界面，研究人员可以通过查询访问私有化统计结果，而无需访问底层原始数据。我们介绍差分隐私和差分隐私汇总统计的构建。然后，我们提出新的算法，用于发布因果效应的差分隐私估计，并生成差分隐私协方差矩阵，从中可以估计任何最小二乘回归。我们通过策展人界面演示了这些方法的应用。
Social scientists often want to analyze data that contains sensitive personal information that must remain private. However, common techniques for data sharing that attempt to preserve privacy either bring great privacy risks or great loss of information. A long literature has shown that anonymization techniques for data releases are generally open to reidentification attacks. Aggregated information can reduce but not prevent this risk, while also reducing the utility of the data to researchers. Even publishing statistical estimates without releasing the data cannot guarantee that no sensitive personal information has been leaked. Differential Privacy, deriving from roots in cryptography, is one formal, mathematical conception of privacy preservation. It brings provable guarantees that any reported result does not reveal information about any one single individual. In this paper we detail the construction of a secure curator interface, by which researchers can have access to privatized statistical results from their queries without gaining any access to the underlying raw data. We introduce differential privacy and the construction of differentially private summary statistics. We then present new algorithms for releasing differentially private estimates of causal effects and the generation of differentially private covariance matrices from which any least squares regression may be estimated. We demonstrate the application of these methods through our curator interface.