A Fuzzy Approach for Privacy Preserving in Data Mining

Abstract

Advances in hardware technology have increased storage and recording capabilities regarding individual’s personal data. Privacy preserving of data has to ensure that individual data publishing will refrain from disclosing sensitive data. Data is anonymized to address the data misuse concerns. Recent techniques have highlighted data mining in ways to ensure privacy. Most anonymization techniques are taken from various fields like data mining, cryptography and information hiding. K-Anonymity is a popular approach where data is transformed to equivalence classes and each class has a set of K- records indistinguishable from each other. But there were many problems with this approach and remedies like ldiversity and t-closeness were proposed to overcome them. This paper addresses the problem of Privacy Preserving in Data Mining by transforming the attributes to fuzzy attributes. Due to fuzzification, exact value cannot be predicted thus maintaining individual privacy, and also better accuracy of mining results were achieved.