Privacy Preserving Data Publishing Based on k-Anonymity by Categorization of Sensitive Values

In many organizations large amount of personal data are collected and analyzed by the data miner for the research purpose. However, the data collected may contain sensitive information which should be kept confidential. The study of Privacypreserving data publishing (PPDP) is focus on removing privacy threats while, at the same time, preserving useful information in the released data for data mining. The number of privacy preserving data publishing techniques is proposed to protect sensitive data from the outside world. K-anonymity is one of the best method which is easy and efficient to achieve privacy in many data publishing applications. It has some weaknesses like data utility reduction and more information loss which need to be focus and optimize. Therefore, the main challenge of research is to minimize the information loss during anonymization process. This paper introduces a new approach for privacy preserving method which is based on categorization of sensitive attribute values. The sensitive attribute value is categorized into high sensitive class and low sensitive class. Anonymization is performed only on those tuples which belong to high sensitive class, whereas tuples belong to low sensitive class published as it is. An experimental result shows that our proposed method is efficient compare to traditional k-anonymity, in terms of data utility and information loss. Index Terms Privacy Preserving; k-Anonymity; Quasi Identifier; Data Utility, Sensitive Classes. ——————————  ——————————