Novel Algorithms-- K-Gen and L-Gen for Implementation of k-Anonymity and l-Diversity Properties

Abstract

Nowadays, the collection of personal data by research organizations and sharing of this data with other organization for business intensions has been increased tremendously. Medical data of individuals is most sensitive among other shared private data. Although some specific values like names and id numbers etc. are removed from shared data to protect the individual privacy the medical data released to research organizations is still susceptible to linking attack which can compromise the patients’ privacy. To prevent linking attack k-Anonymity Property is used. In a k-anonymized dataset, each record is indistinguishable from at least k−1 other records with respect to certain ―identifying‖ attributes. As k-anonymity cannot prevent attribute disclosure, to go beyond k-anonymity the notion of l-diversity is used to address this; l-diversity requires that each equivalence class has at least l well-represented values for each sensitive attribute. In this paper we implemented a system using VB.Net for k-anonymous and distinct l-diverse table generation. We propose two algorithms K-Gen and L-Gen for the system implementation with generalization index values. The generalization index values controls the levels of generalization for each ―identifying‖ attribute. Experimental results show that the proposed algorithms have lesser Discernibility value and comparable information loss as compared with existing methods. The system implemented in this paper protects patients’ identity by efficient implementation of k-anonymity and distinct l-diversity properties, providing faster data release with intuitive GUI. It also offers a cost-effective solution to patients’ data holding organizations, as they need not buy any proprietary black boxes to protect the released data.