Research project MTM2006-05550, Spanish Ministry of Science and Education.
October 2006-September 2009.
Members
Introduction and objectives
The current information society requires continuously to manage a larger amount of data for decision making. Among the main agents devoted to the dissemination of aggregated or tabular data we find the National Statistical Agencies (NSAs). However they must preserve the individual rights of privacy and data confidentiality. Data protection techniques are the tools used by NSAs for safe dissemination of information. Several European NSAs are considering a new strategy for data protection, the minimum-distance controlled tabular adjustment, which was suggested by this research group in a previous research project. This technique reduces to a linear or quadratic optimization problem, depending on the distance to be used, which in practice turned out to be a challenge for nonpolynomial methods such as the simplex. Preliminary results with small and medium size tables showed that interior-point methods (in which the research group has a wide experience) are more effective, being up to 60 times faster than the simplex in some instances. However, for real tables, of millions of cells and hundreds of thousands of constraints, general interior-point algorithms are even inefficient, and we need specialized ones. The objective of the project is the development of interior-point algorithms able to efficiently solve massive tabular data. For this purpose, among others tasks:
Click here for a list of papers and conference presentations generated
within the project.