基于Logistic模型的差分隐私推荐算法  

Differential Privacy Recommendation Based on Logistic Model

作　　者：邓雷升[1] 杨竞 DENG Leisheng;YANG Jing(No.30 Institute of CETC,Chengdu Sichuan 610041,China)

机构地区：[1]中国电子科技集团公司第三十研究所,四川成都610041

出　　处：《通信技术》2024年第11期1190-1194,共5页Communications Technology

摘　　要：针对大数据环境下的用户隐私保护问题,提出一种基于Logistic模型的差分隐私推荐算法。首先,改进了差分隐私保护下的系统模型,用户对自身数据进行编码后发送给可信的服务器进行数据加扰处理,可信的服务器迭代处理后再推送给分析服务器进行推荐。系统确保了用户的数据只能被可信的服务器进行搜集整理,不会泄露用户隐私。其次,基于Logistic模型给出了差分隐私推荐算法,该算法将用户的数据作为一个整体进行分析,提高了效率。To address the problem of user privacy protection in big data environment,this paper proposes a differential privacy recommendation algorithm based on Logistic model.First,it improves the system model under differential privacy protection,where users encode their own data and send them to a trusted server for data scrambling.Then,the trusted server carries out iterative processing and pushes them to the analytics server for recommendation.The system ensures that user’s data can only be collected and processed by trusted servers,thus not disclose user privacy.Finally,based on Logistic model,a differential privacy recommendation algorithm is given,which analyzes the user’s data as a whole and improves the efficiency.

关 键 词：压缩降维 LOGISTIC模型 差分隐私 统计查询 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]