Privacy-preserving sensitive data publishing in case of outliners challenges

Abstract

The purpose of keeping personal data is to prevent malicious individuals from hacking or violating privacy in order to extort or take advantage of the data owner or the person in charge of the data. Outliers present significant challenges in machine learning and data security due to their substantial deviation from the norm, making them attractive targets for potential security breaches. Outliers attract the attention of hackers due to their potential to reveal personally sensitive information. This paper addresses the outlier problem within data privacy concerns, focusing specifically on personal data that includes outliers. A novel framework, namely Handling Outlier Anomaly Privacy Violation (HOPV), designed for implementation on data controllers' web servers, is proposed in order to monitor and mitigate this issue. The HOPV framework incorporates an advanced outlier detection algorithm complemented by sophisticated data generalization and Laplace Mechanism Perturbation techniques. Empirical results elucidate the remarkable performance superiority of our software module when juxtaposed with existing products in the field.