基于本地化差分隐私的政务数据共享隐私保护算法研究

Research on Privacy Protection Algorithm for Government Data Sharing Based on Local Differential Privacy

[目的/意义]为了合理化决策,通常一个政府部门会根据业务需求向其他部门共享某类数据,为本部门管理或服务决策提供辅助参考依据。数据共享在其中至关重要,但若在没有适当预防措施的情况下就共享政务数据,将容易造成隐私信息的泄露。[方法/过程]针对政府部门间共享统计数据的场景,提出一种基于本地化差分隐私的政务数据共享方法。该方法在算法Generalized randomized response(GRR)的基础上引入数据分箱思想,通过等宽分箱将数据记录分入更小的数据域范围内,以克服当前隐私保护算法在数据域较大且数据量较少时统计误差大的问题。[结果/结论]将所提算法与GRR算法在仿真数据集和真实数据集上均进行了对比分析,实验结果表明该算法可有效降低统计误差,并能在不同分布和数据域大小下保持其效用性。

[Purpose/Significance]In order to rationalize the decision-making,usually a government department will share some kind of data with other departments according to the business needs,and provide auxiliary reference for the management or service decision-making of the department.Data sharing is of utmost importance.However,if government data is shared without proper precautions,private information could easily be leaked.[Method/Process]Aiming at the scenario of sharing statistical data among government departments,a government data sharing approach based on local differential privacy is proposed.This approach introduces the idea of data binning based on the Generalized Randomized Response(GRR)algorithm.Equi-width binning is adopted to divide the data into smaller data domains to overcome the problem of large statistical errors in the current privacy protection algorithms when the data domain size is large and the amount of data is small.[Result/Conclusion]We compared the proposed algorithm with the GRR on both the simulation datasets and a real dataset.Experimental results show that the proposed algorithm can effectively reduce statistical errors and maintain its utility under different distributions and data domain sizes.