Affiliation:
1. Department of Geography The Ohio State University Columbus Ohio USA
2. Center for Spatial Data Science University of Chicago Chicago Illinois USA
Abstract
Privacy and utility are two important objectives to consider when releasing census data. However, these two objectives are often conflicting, as protecting privacy usually necessitates introducing noise into the data, which compromises data utility. Determining the appropriate level of privacy protection presents a significant challenge in the data release. Therefore, it is necessary to investigate the tradeoff between privacy and utility before making a final decision on the level of privacy protection. In this article, we propose a multiobjective optimization framework to generate multiple optimal solutions that satisfy the two objectives of privacy and utility, as well as to analyze the tradeoff between privacy and utility for decision‐making. This framework relocates individuals susceptible to revealing their identities to protect their privacy. We maximize the number of individuals relocated while maximizing the utility of the data after relocations. The proposed framework is tested using synthetic population data in Franklin County, Ohio. Our experimental results show that the framework can efficiently generate a collection of optimal solutions and can be used to effectively balance privacy and utility.
Subject
Earth-Surface Processes,Geography, Planning and Development
Reference54 articles.
1. The U.S. Census Bureau Adopts Differential Privacy
2. Confidentiality Protection in the 2020 US Census of Population and Housing;Abowd J. M.;Annual Review of Statistics and Its Application,2023
3. The 2020 Census Disclosure Avoidance System TopDown Algorithm;Abowd J. M.;Harvard Data Science Review,2022
4. Optimizing Watchtower Locations for Forest Fire Monitoring Using Location Models;Bao S.;Fire Safety Journal,2015
5. Disclosure Control of Microdata;Bethlehem J. G.;Journal of the American Statistical Association,1990