Statistics Professor Says Databases Must Balance Privacy, Utility Carnegie Mellon News (08/30/07) Potts, Jonathan
as it appeared in the August 31, 2007 edition of ACM TechNews.
Carnegie Mellon University statistics professor George Duncan says that organizations with large databases such as the U.S. Census Bureau, which collects tremendous amounts of personal information, need to find ways to protect individuals' privacy while making the data available to researchers. Duncan believes that traditional methods of "de-identifying" records such as removing Social Security numbers and birth dates do not adequately protect sensitive information because if someone knows enough about the data they could use other characteristics to identify individuals. Unfortunately, the information that can be used to re-identify records is often the information that is most useful to the researchers. "The question is, 'How can data be made useful for research purposes without compromising the confidentiality of those who provided the data?'" Duncan asks. Possible solutions include establishing administrative procedures that restrict data access to approved personnel, implementing restrictions on the use of information, and developing statistical methods that de-identify records so that users cannot readily reconstruct personal identities but researchers can still view the required information. "Achieving 'adequate' privacy will require engineering innovation, managerial commitment, information cooperation of data subjects, and social controls," Duncan wrote in a commentary published in the journal Science. ACM's Public Policy Committee (USACM) provided testimony on protecting Social Security numbers at a recent Congressional hearing. For more information, go to http://www.acm.org/public-policy/public-policy-1?pageIndex=1 Click Here to View Full Article
