Enhanced utility in preserving privacy for multiple heterogeneous sensitive attributes using correlation and personal sensitivity flags
K. Ashoka, Poornima B H · 2017
Preserving privacy is becoming a key apprehension as personal data is publicly available in recent years. Most of the present Privacy Preserving Data Publishing (PPDP) methods could not process multiple, heterogeneous sensitive attributes with different levels of sensitivity requirements. This motivates us to suggest a novel methodology that can handle multiple heterogeneous (both numerical and categorical) sensitive attributes. Unlike the existing methods that applies same amount of privacy for all the data records without considering record owner's concrete needs, our technique considers Personal Sensitivity Rating Flags (PSRF) from record owners and apply privacy preserving techniques on those records that actually needs it. Also the correlation among the attributes is calculated and based on this correlation, highly correlated attributes are generalized as minimum as possible by generalizing other attributes more. Therefore our approach will preserve sufficient privacy by retaining maximum amount of information from original data. Our proposed technique is theoretically analyzed and mathematical analysis surpasses earlier works with sufficient experiments.