Authors : S. Balamurugan and P. Visalakshi
Abstract: Publishing microdata amplifies the problem arising out of individual privacy entity. This study investigates the problem of privacy preservation hazard of mined Conditional Functional Dependency (CFD) against (d, l) Inference Model using CFPGBS methods. The major problem of the above mentioned methods is that, it protects privacy only for the entire table without considering the attribute wise privacy for both CFDs and FFDs against the (d, l) Inference Model. In order to overcome these limitations, Boyce-Codd Normal Form (BCNF) method has been presented to facilitate publishing of multiple subtables and it is anonymized through (d, l) Inference Model however the integration of different published subtables also needs to guarantee privacy rules for CFDs. The construction of the initial partitions for the driven bottom-up approach is performed through Fuzzy Binomial Distribution (FBD). Experimental results show that the proposed BCNF (d, l) Inference Model can adeptly anonymize the microdata with less information loss as compared to (d, l) Inference Model with CFD. The effectiveness and privacy results of proposed BCNF (d, l) Inference Model with FBD is also significant in comparison to the existing Inference Model with Log-Skew-Normal Alpha-Power distribution (LSKNAPD) function.
S. Balamurugan and P. Visalakshi, 2016. Boyce-Codd Normal Form (BCNF) Based Privacy-Preserving Publishing Multiple Subtables with Conditional Functional Dependencies. Asian Journal of Information Technology, 15: 347-358.