PSI (ΨΨ\Psi): a Private data Sharing Interface (working paper)
Marco Gaboardi Department of Computer Science and Engineering, University at Buffalo, SUNY. Work done in part while at the University of Dundee, UK and visiting the Center for Research on Computation、Society, John A. Paulson School of Engineering、Applied Sciences, Harvard University. James Honaker http://hona.kr Gary King Albert J. Weatherhead III University Professor, Harvard University, Institute for Quantitative Social Science. http://GaryKing.org Jack Murtagh Center for Research on Computation、Society, John A. Paulson School of Engineering、Applied Sciences, Harvard University. Kobbi Nissim Department of Computer Science, Georgetown University, and Center for Research on Computation、Society, John A. Paulson School of Engineering、Applied Sciences, Harvard University. Jonathan Ullman College of Computer and Information Sciences, Northeastern University. Work done in part while affiliated with the Center for Research on Computation、Society, John A. Paulson School of Engineering、Applied Sciences, Harvard University. Salil Vadhan Center for Research on Computation、Society, John A. Paulson School of Engineering、Applied Sciences, Harvard University. Work done in part while visiting the Shing-Tung Yau Center and the Department of Applied Mathematics at National Chiao-Tung University in Taiwan. Also supported by a Simons Investigator Award. with contributions from Nabib Ahmed Andreea Antuca Brendan Avent Jordan Awan Christian Baehr Connor Bain Victor Balcer Thomas Brawner Jessica Bu Mark Bun Stephen Chong Fanny Chow Katie Clayton Holly Cunningham Vito D’Orazio Gian Pietro Farina Anna Gavrilman Benjamin Glass Caper Gooden Paul Handorff Raquel Hill Alyssa Hu Jason Huang Justin Kaashoek Allyson Kaminsky Chan Kang Murat Kuntarcioglu Vishesh Karwa George Kellaris Michael Lackner Jack Landry Hyun Woo Lim Giovanni Malloy Michael Lopiccolo Nathan Manohar Ross Mawhorter Dan Muise Marcelo Novaes Ana Luisa Oaxaca Raman Prasad Sofya Raskhodnikova Grace Rehaut Ryan Rogers Or Sheffet Adam D. Smith Thomas Steinke Kathryn Taylor Julia Vasile Clara Wang Haoqing Wang Remy Wang Lancelot Wathieu David Xiao Anton Xue and Joy Zheng
Abstract
We provide an overview of the design of PSI (“a Private data Sharing Interface”), a system we are developing to enable researchers in the social sciences and other fields to share and explore privacy-sensitive datasets with the strong privacy protections of differential privacy.
中文速览
社会科学研究者越来越被要求公开数据集，但许多涉及个人隐私的数据集要么通过传统"去标识化"手段处理后仍存在重识别风险，要么因保护措施繁琐而完全无法获取，严重阻碍了研究的推进。为此，研究团队基于差分隐私（differential privacy）技术，设计并开发了 PSI（Private data Sharing Interface）系统，让数据存储者和分析者无需具备隐私或统计专业知识，就能通过友好界面对敏感数据集进行探索性统计分析，同时获得严格的数学隐私保证。该系统与 Dataverse 等主流数据仓库及 Zelig、TwoRavens 等社会科学常用分析工具深度集成，允许研究者在正式申请数据访问权限之前，先通过加噪统计结果判断数据集是否符合研究需求，从而减少大量无效的审批流程。PSI 的意义在于，它为隐私保护与数据开放之间提供了一条可操作的中间路径，有望大幅降低敏感数据共享的门槛，推动社会科学等领域的研究复现与协作。
原文 arXiv:1609.04340;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1609.04340v3