Òòööð ¹¹òóòýññþþøøóò × Ààöö

Adam Meyerson 1 Ryan Williams2 Mar h 2003 CMU-CS-03-113 S hool of Computer S ien e Carnegie Mellon University Pittsburgh, PA 15213 Abstra t The strategy of k-anonymization, proposed by Sweeney [7℄ is a te hnique for prote ting the priva y of individuals while allowing large-s ale data mining. An optimum anonymization is one whi h obtains the required priva y prote tion (insuring that every individual is a member of a group of k identi al individuals) while minimizing the amount of data hidden from the data miner. We prove that nding su h an optimum anonymization is NP-Hard.