Efficient anonymization algorithm for multiple sensitive attributes

Abstract

The data of medical applications over the internet contains sensitive data. There exist several methods that provide privacy for these data. Most of the privacy-preserving data mining methods make the assumption of the separation of quasi-identifiers (QID) from multiple sensitive attributes. But in reality, the attributes in a dataset possess both the features of QIDs and sensitive data. In this paper privacy model namely (vi…vj)-diversity is proposed. The proposed anonymization algorithm works for databases containing numerous sensitive QIDs. The real dataset is used for performance evaluation. Our system reduced the information loss for even huge number of attributes and the values of sensitive QID’s are protected.