The growing need for more and more detailed statistical information has an increased impact on the confidentiality issues. The increased computing power of the end-users has led to greater threats of disclosure of confidential information. This project aims at the enhancement of tools and the development of new methods for disclosure limitation. New software tools for both the protection of microdata and tabular data will be developed. The research is aimed primarily at new methods for releasing business microdata while new optimisation methods will be developed for the protection of hierarchical tables.
The research and development of new techniques for Statistical Disclosure Control is the main objective of this project. The need for more statistical information has a consequence that we should pay more attention to the confidentiality aspects. The impact on the Statistical Offices is very big if we fail to solve this problem adequately. The project will aim at the development of practical tools as well as new research to support the development of these tools. Attention will be paid to both tabular data as well as microdata. The development of these standard tools serves the harmonisation of the statistical production in Europe and will enforce the leading role of Europe in this topic.
The work in this project can be divided in two main streams, which is reflected in the composition of the project team.
The team includes both NSI's and universities:
1. The disclosure protection of microdata. In the field of microdata several new techniques will be investigated and also implemented in m-ARGUS. The traditional techniques (global recoding and local suppression) serve quite well for microdata on individuals, but are inadequate for business microdata. New methodologies like Masking techniques, Post-randomisation (PRAM), micro-aggregation and noise-addition will be investigated and evaluated. Implementation in m-ARGUS will allow for experimenting with these techniques. To measure the quality of the methods applied; disclosure risk and information loss models will be implemented too.
2.The disclosure protection of tabular data for tabular data the emphasis is on the complexity of the tables. Hierarchical and linked tables imply much more complex optimisation problems to be solved. Two approaches for finding the optimal solution will be developed. Besides these some approximation methods will be implemented and their effects will be investigated. We will pay attention to the testing of this new methodology, investigate re-identification risks. The impacts on the analytical power of the protected datasets will be studied and we will ensure the exploitation of the results by presenting the results on relevant meetings and making them available on the WEB.
This consortium aims at the development of practical tools for Statistical Disclosure Control (SDC). The milestones will be the development of major extensions to the ARGUS-software. For these developments many new research papers will be needed and produced. Besides research in SDC research in numerical optimisation is required and included. The major extensions of the ARGUS software will be the production of safe business microdata and safe hierarchical and linked tables.
Funding SchemeCSC - Cost-sharing contracts
SW1V 2QQ London
M13 9PL Manchester
38207 Santa Cruz De Tenerife
PL4 8AA Plymouth
SO17 1BJ Southampton