Solving the sensitive itemset hiding problem whilst minimizing side effects on a sanitized database

Abstract

Mining frequent itemsets from huge amounts of data is an important issue in data mining, with the retrieved information often being commercially valuable. However, some sensitive itemsets have to be hidden in the database due to privacy or security concerns. This study aimed to secure sensitive information contained in patterns extracted during association-rule mining. The proposed approach successfully hides sensitive itemsets whilst minimizing the impact of the sanitization process on nonsensitive itemsets. Our approach ensures that any modification to the database is controlled according to its impact on the sanitized database. The results of simulations demonstrate the benefits of our approach. © 2011 Springer-Verlag.