Cluster Based Attribute Slicing: A New Approach for Privacy Preservation

Various anonymization techniques have been proposed for publishing a microdata. Anonymization techniques hide the sensitive data from the attackers. Examples of these techniques are slicing, bucketization and generalization. Generalization hides sensitive data but it loses lot of data. Bucketization make the difficult to detect sensitive attribute by randomizing sensitive attribute but it does not prevent relation between them. Slicing is better technique amongst all remaining technique because it cannot lose information and it is used for maintain the relation between attributes. In case of slicing, equal-width discretization is used to convert continuous attribute into categorical attribute. Equal-width discretization has very time consuming technique. To solve this problem we propose cluster based attribute slicing algorithm. Proposed technique does not take more to sort a data.