Subscribe

Issue No.02 - April-June (2008 vol.5)

pp: 87-98

ABSTRACT

The fundamental problem for inference control in data cubes is how to efficiently calculate the lower and upper bounds for each cell value given the aggregations of cell values over multiple dimensions. In this paper, we provide the first practical solution for estimating exact bounds in two-dimensional irregular data cubes (i.e., data cubes in which certain cell values are known to a snooper). Our results imply that the exact bounds cannot be obtained by a direct application of the Fr\\'{e}chet bounds in some cases. We then propose a new approach to improve the classic Fr\\'{e}chet bounds for any high-dimensional data cube in the most general case. The proposed approach improves upon the Fr\\'{e}chet bounds in the sense that it gives bounds that are at least as tight as those computed by Fr\\'{e}chet, yet is simpler in terms of time complexity. Based on our solutions to the fundamental problem, we discuss various security applications such as privacy protection of released data, fine-grained access control and auditing, and identify some future research directions.

INDEX TERMS

Inference engines, Data dependencies

CITATION

Haibing Lu, Yingjiu Li, "Practical Inference Control for Data Cubes",

*IEEE Transactions on Dependable and Secure Computing*, vol.5, no. 2, pp. 87-98, April-June 2008, doi:10.1109/TDSC.2007.70217REFERENCES