A New Approach to Utility-Based Privacy Preserving in Data Publishing

Abstract

A fundamental problem in privacy-preserving data publishing is how to make the right trade-off between privacy risks and data utility. Anonymization techniques are used both to reduce privacy risks and to create anonymized dataset. The anonymized dataset can be grouped together into equivalence classes. The Equivalence Classes are classified into two groups based on the utility provided to the data recipients: Utility Equivalence Class (UEC) and Outlier Equivalence Class (OEC). The OEC contains records that have been fully suppressed by anonymization techniques resulting in no data utility. In this study, a new approach is proposed by reducing the number of outlier records in order to increase the data utility. In the proposed model, k-anonymity and l-diversity privacy models are used together to reduce the privacy risks. The Average Equivalence Class Size is used in measuring the data utility. According to the experimental results, the data utility is increased with the use of our proposed model while keeping the delicate balance between privacy risks and data usefulness.