Thirty years of optimization-based SDC methods for tabular data
Jordi Castro(a),(*)
Transactions on Data Privacy 16:1 (2023) 3 - 13
Abstract, PDF
(a) Dept. of Statistics and Operations Research, Universitat Politècnica de Catalunya, Jordi Girona 1-3, 08034, Barcelona, Catalonia.
e-mail:jordi.castro @upc.edu
|
Abstract
In 1966 Bacharach published in Management Science a work on matrix rounding problems in two-way tables of economic statistics, formulated as a network optimization problem. This is likely the first application of optimization/operations research for statistical disclosure control (SDC) in tabular data. Years later, in 1982, Cox and Ernst used the same approach in a work in INFOR for a similar problem: controlled rounding. And thirty years ago, in 1992, a paper by Kelly, Golden and Assad appeared in Networks about the solution of the cell suppression problem, also using network optimization. Cell suppression was used for years as the main SDC technique for tabular data, and it was an active field of research which resulted in several lines of work and many publications. The above are some of the seminal works on the use of optimization methods for SDC when releasing tabular data. This paper discusses some of the research done this field since then, with a focus on the approaches that were of practical use. It also discusses their pros and cons compared to recent techniques that are not based on optimization methods.
|