Advances in Multimedia, Software Engineering and Computing Vol.2

Abstract

The database quality hidden after rules is always judged by data user. Only high quality to data users is real high quality. In many privacy-preserving association rule mining algorithms, there is no accurate quantitative assessment on effect of hiding on database. To address the problem, the paper presented a kind of personalized privacy-preserving association rule mining algorithm that has less damage to database from view of data user. The algorithm selects data to be deleted according to importance index expensed by each support number of each restrictive rule. In order to evaluate algorithm performance, a standard to measure damage survived from database cleaning oriented to users was also provided. © 2011 Springer-Verlag Berlin Heidelberg.