FuzzyDP: Fuzzy-based big data publishing against inquiry attacks
Youyang Qu, Shui Yu, Longxiang Gao, Sancheng Peng, Yong Xiang, Liang Xiao · 2017
Privacy issues are confronting rapidly increasing challenges in this big data era. There is an increasing trend that more adversaries and hackers are aiming at individual privacy with updated technology, which leads to financial loss and safety issues. We have an observation that existing models, for example, classic differential privacy, provide certain privacy protection to statistical databases with unsatisfying data utility. Motivated by this, we propose FuzzyDP model using fuzzy logic with e-differential privacy, which is the early work using fuzzy logic to cluster datasets in differential privacy. By fuzzifying the datasets into clusters, we can reduce the global sensitivity and therefore control the noise in a satisfying level. In this way, the proposed model achieves better data utility than classic laplace mechanism, while maintains the same privacy level. Our solid theoretical analysis and extensive experiments prove the effectiveness of the proposed model.