Privacy Problems with Anonymized Transaction Databases
In: Discovery Science 2004, 2-5 Oct 2004, Padova, Italy.
In this paper we consider privacy problems with anonymized transaction databases, i.e., transaction databases where the items are renamed in order to hide sensitive information. In particular, we show how an anonymized transaction database can be deanonymized using non-anonymized frequent itemsets. We describe how the problem can be formulated as an integer programming task, study the computational complexity of the problem, discuss how the computations could be done more efficiently in practice and experimentally examine the feasibility of the proposed approach.