Towards privacy-anomaly detection: Discovering correlation between privacy and security-anomalies

Abstract

In this paper a notion of privacy-anomaly detection is presented where normative privacy is modelled using k-anonymity. Based on the model, normative privacy-profiles are constructed, and deviation from normative privacy-profile at runtime is labelled as a privacy-anomaly. Furthermore, the paper investigates whether there is a correlation between security-anomalies and privacy-anomalies, that is, whether the privacy-anomalies labelled by privacy-anomaly detection system are detected by conventional security-anomaly detection system used for detecting malicious accesses to databases by insiders.