一种分布式事务数据的差分隐私发布策略

目前隐私保护的事务数据发布研究多是基于集中式结构.针对分布式结构下事务数据发布问题,为保护数据隐私,同时最大化数据效用,提出一种满足差分隐私约束的发布策略.首先,将结果效用性优化与差分隐私约束相结合,构建分布式非线性规划模型.然后,基于全局与局部数据设计两种解决方案安全求解该分布式模型.理论分析与实验结果均表明,所提出的发布策略是安全的且满足差分隐私要求,具有很好的实用性.

Differential Privacy Publishing Strategy for Distributed Transaction Data

In the research of privacy preserving transaction data publishing, the existing methods are always designed for the centralized structure. The paper proposes a differential privacy publishing strategy to protect data privacy and maximize utility of the output data in the distributed environment. The new method combines the utility optimization of the output with differential privacy constraints and builds a distributed nonlinear programming model. Furthermore, two solutions based on global and local data respectively are designed to solve the distributed model securely. As shown in the theoretical analysis and the experimental results, the publishing strategy can achieve significant improvements in terms of privacy, security, and applicability.