基于本地差分隐私的多维数据特征信息查询算法研究

钟晓书1; 杨勇2

基于本地差分隐私的多维数据特征信息查询算法研究

引用本文：钟晓书1，杨勇2.基于本地差分隐私的多维数据特征信息查询算法研究[J].计算技术与自动化,2026,(2):23-28

摘要点击次数: 56

全文下载次数: 21

作者	单位
钟晓书1，杨勇2	（1. 陕西工业职业技术大学,陕西咸阳 712000；2.陕西省大数据集团有限公司,陕西西安 710061）

中文摘要:针对云存储环境下多维数据查询面临的隐私保护不足与大规模数据处理效率低等问题，提出了一种基于本地差分隐私的多维数据特征信息查询算法。该算法根据数据敏感度分布设计动态噪声方案，实现分段或逐点差分处理，在保护隐私的同时保持数据整体统计特征；结合R树空间索引结构与主成分分析，优化多维数据查询效率并增强个体隐私防护能力。基于UCI数据集中股票市值变化数据的实验结果表明，本算法在皮尔逊相关系数等多项指标上均优于对比方法，数据查询保留度最高达95%，有效平衡了隐私保护与数据可用性，为云存储环境下的隐私安全数据分析提供了新的技术路径。

中文关键词:本地差分隐私多维数据拉普拉斯分布查询算法

Research on Feature Information Query Algorithm of Multi-dimensional Data Based on Local Differential Privacy

Abstract:In response to the issues of insufficient privacy protection and low processing efficiency of large-scale data in cloud storage environments for multi-dimensional data querying, propose a multi-dimensional data feature information query algorithm based on local differential privacy. This algorithm designs a dynamic noise scheme according to the distribution of data sensitivity, achieving segmented or point-by-point differential processing, while protecting privacy while maintaining the overall statistical characteristics of the data. Combined with the R-tree spatial index structure and principal component analysis, it optimizes the efficiency of multi-dimensional data querying and enhances the individual privacy protection capability. Experimental results based on the stock market value change data from the UCI dataset show that this algorithm outperforms the comparison methods in multiple indicators such as Pearson correlation coefficient, with the data query retention rate reaching up to 95%. It effectively balances privacy protection and data availability, providing a new technical path for privacy security data analysis in cloud storage environments.

keywords:local differential privacy multi-dimensional data Laplace distribution query algorithm

查看全文 查看/发表评论 下载pdf阅读器