Slicing: A New Approach to Privacy Preserving Data Publishing

In the last decade, anonymization techniques such as generalisation, bucketization and slicing have appeared in privacy-preserving data publishing. Protecting identity privacy means using generalisation, but thisl often lowers the usefulness of the data, particularly with datasets containing many features. Although useful for privacy, bucketization can still lead to membership disclosure and becomes less reliable when it’s hard to separate quasi-identifying from sensitive attributes. For dealing with these restrictions, a new method referred to as slicing is now introduced. Both horizontal and vertical partitioning of data save moreuseful information than if the data is generalised and also keep information about membership private. Besides, slicing is equipped to handle lots of data features and matches ℓ-diversity standards for confidentiality. A good algorithm for doing sliced data lines up with privacy rules. We found that slice units work better than generalised units at keeping the importance of data when dealing with sensitive attributes. In addition, the use of slicing ensures that no connexion can reveal membership. This work adds the t-closeness slicing (TCS) method which helps secure transactional data, paying special attention to issues of membership privacy, identity privacy and attribute privacy. On large datasets, TCS efficiently works using the time complexity O(n log n). The results from three different transactional datasets demonstrate that TCS improves privacy and gives better value from data than other approaches