基于差分隐私的贝叶斯网络隐私保护算法的改进研究

Research on Improvement of Bayesian Network Privacy Protection Algorithm Based on Differential Privacy

针对数据企业对用户信息以及开放数据趋势下政府数据发布工作对脱敏保护算法的迫切需求,文章提出一种基于差分隐私保护理论的具有属性段首选机制和基于聚类算法的贝叶斯网络改进型算法FCPrivBayes。该算法避免了对首个属性段属性的随机化选择,并用聚类的方法取代等宽法对数据进行离散化处理。实验数据表明,在保障数据隐私的前提下,FCPrivBayes有效提升了数据的可用性指标,为企业保护数据、政府发布数据提供了新的技术方案,有利于用户隐私保护工作的推进和大数据产业的发展。

In response to the urgent need for desensitization protection algorithms by the data companies and open government publishing data,under the strict differential privacy theory,an improved Bayesian network algorithm FCPrivBayes with an attribute segment preference mechanism and a clustering algorithm is proposed,which avoids the random selection of the attributes of the first attribute segment,and uses the clustering method to replace the equal-width method to discretize the data.Experimental data show that FCPrivBayes effectively improves data utility indicators while ensuring the data privacy protection effect.Which provides new technical options for data companies to protect data and for government to release data,and benefits the user privacy protection and the development of the big data industry.