|Títol||Aggregation methods to evaluate multiple protected versions of the same confidential data set|
|Publication Type||Book Chapter|
|Year of Publication||2002|
|Authors||Valls A, Torra V, Domingo-Ferrer J|
|Editor||.Grzegorzewski P, Hryniewicz O., Gil M.A|
|Book Title||Advances in Soft Computing|
This work is about disclosure risk for national statistical offices and, more particularly, for the case of releasing multiple protected versions of the same micro-data files. This is, several copies of a single original data file are released to several data users. Each user receives a protected copy, and the masking method for each copy is selected according to the research interests of the user: the selected masking method is such that it minimizes the information loss for his/her particular research. Nevertheless, multiple releases of the same data increase the disclosure risk. This is so, because coalitions of data users can reconstruct original data and, thus, find the original (non-masked) information. In this work we propose a tool for evaluating this reconstruction.
- Quant a IIIA