Authors : S. Balamurugan and P. Vislakshi
Abstract: Publishing individual data has grabbed more attention towards individual privacy in recent days. It has become an active research area and several works have been carried out to preserve privacy. The (d, l) Inference Model each and every group contains l sensitive values with frequency similarity value and which is controlled by the parameter d. Recent researches do not focus on CFD against the (d, l) Inference Model and doesn’t work with a Frequency Distribution Method for initial partition phase. In order to overcome these two problems, this study proposes a novel (d, l) Inference Model to deal with adversarial information rendered by Conditional Functional Dependencies (CFDs). However, discovering the quality of CFD is a challenging task. In order to solve the above mentioned problem, Automatic Compact Frequent Pattern Growth Branch Sort algorithm (CFPGBS) is proposed for mining the best CFD patterns and removing the less quality CFD patterns. A compact pattern tree is developed, that captures CFD patterns information with insertion phase and provides the better pattern mining performance for CFD patterns. The construction of the initial partitions for the bottom-up approach driven is performed by Log-Skew-Normal Alpha-Power distribution (LSKNAPD) frequency distribution function. Experimental results show that the proposed (d, l) Inference Model can proficiently anonymize the micro data with a low information loss against CFD.
S. Balamurugan and P. Vislakshi, 2016. Proposing New Strategy for Privacy Preserving Microdata Publishing with Conditional Functional Dependencies. Asian Journal of Information Technology, 15: 298-311.