Privacy protection

A recent article (1) describes how an individual's privacy can be assured in a database with a restriction on aggregate queries that are allowed. In a database allowing aggregate cross-sectional queries, information at the element level can be revealed by sub segmentation. The suggested solution provides bounds on queries that return subsets with individual information revelation.

Research in this area has been lagging. Most of the effort is currently expended in the collection, aggregation and reporting of information without sufficient concern for the privacy of such information. There are two fundamental avenues of research in this general area. First, one has to create a mathematical foundation for the determination of the minimum amount of data at the individual level (to be stored) that is necessary and sufficient to complete the task at hand and second, one has to determine decision quality or the ability to use all available information in the decision process as a function of stored data. There have been two weaknesses in this process. There is a tendency to collect whatever data are available first and ask questions later as the cost of collection and storage decline fast (2) and most of the stored information is never used in the decision process. This has created problems not only for reaching better decisions but also in the protection afforded. This is because, highly segmented datasets provide less protection against accidental revelation of information as well as planned ones.

Better designs are needed in the definition of necessary and sufficient data, decision processes that operate on such data to improve decision quality and databases that automatically prevent information revelation.

(1) Protecting confidential data with math. Published: Friday, December 16, 2011 - 12:37 in Mathematics & Economics. Source: Society for Industrial and Applied Mathematics

(2) Flexibility: Flexible Companies for the Uncertain World. http://www.routledge.com/books/details/9781439816325/