Team Receives Grant for Data-Privacy Research

A team of researchers at Penn State University, in collaboration with researchers from Cornell University and Carnegie Mellon University, has been awarded a $2 million grant from the National Science Foundation Cyber-Enabled Innovation and Discovery program for a research project, "CDI-Type II: Collaborative Research: Integrating Statistical and Computational Approaches to Privacy." The team, led by Aleksandra Slavkovic, associate professor of statistics, and Adam Smith and Sofya Raskhodnikova, faculty members in the Department of Computer Science and Engineering, will use a multidisciplinary approach combining statistics and computer science to study how research data in the social and heath sciences can be made more widely available without compromising privacy. The project also will increase awareness of data-privacy issues in three academic communities—computer science, statistics and the social sciences—and will promote further research on statistical methods to limit disclosure of sensitive information, on cryptography, and on privacy-preserving data mining.

Slavkovic and the other members of the research team believe that data privacy is a fundamental problem of the modern information infrastructure. Enormous volumes of personal and sensitive data are collected and archived by many organizations—for example, health networks, government agencies, search engines, and social-networking websites. "Opening up these databases to researchers has tremendous potential benefits for society as whole. Still, the release of information from sensitive data repositories can be devastating to the privacy of individuals and organizations," the team reported. Slavkovic said that "the challenge is to maintain the usefulness of these databases while limiting the possibility of accidental leaks or privacy breaches."

The research team plans to develop new methodologies for private statistical analyses that will enable government agencies and social-science and public-health researchers to share and publish their data under stronger privacy guarantees. The team also aims to develop an understanding of the practical potential of the techniques they develop by applying them to social-science data in collaboration with such entities as the U.S. Census Bureau, the Interuniversity Consortium for Political and Social Research (ICPSR), and Microsoft.