北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2023, Vol. 46 ›› Issue (3): 115-120.

• 论文 • 上一篇    

满足差分隐私的逻辑回归矩阵分解推荐算法

杜茂康1,彭俊杰1,胡勇进2,肖玲3   

  1. 1. 重庆邮电大学 数据科学与复杂性系统管理重点实验室 2. 重庆邮电大学 计算机科学与技术学院 3. 徐州工程学院 数学与统计学院

  • 收稿日期:2022-03-09 修回日期:2022-07-04 出版日期:2023-06-28 发布日期:2023-06-05
  • 通讯作者: 杜茂康 E-mail:dumk@cqupt.edu.cn
  • 基金资助:

    重庆市自然科学基金项目(cstc2021jcyj-msxmX0557)

Logistic Regression Matrix Factorization Recommendation Algorithm for Differential Privacy

DU Maokang1PENG Junjie1HU Yongjin2Xiao Ling3   

  • Received:2022-03-09 Revised:2022-07-04 Online:2023-06-28 Published:2023-06-05

摘要:

为了提高隐私保护下的推荐算法准确性,提出了一种满足差分隐私保护的逻辑回归矩阵分解推荐算法。该算法首先将隐式数据的矩阵分解转换为分类问题并以概率方式对其建模;然后采用sigmoid函数对预测评分进行非线性变换,将原始的矩阵分解问题转换成用户隐因子和项目隐因子的优化问题,并对目标函数添加随机噪音进行扰动,使算法满足差分隐私保护。在Movielens100K,Movielens1M和YahooMusic数据集上进行实验,并与现有算法进行对比,该算法在F1值指标上分别提升了9.29%,7.40%和3.61%。理论分析和实验结果表明,所提算法在实现用户隐式反馈数据保护的同时还能有效地保证推荐结果的准确性,具有良好的应用价值。

关键词: 隐式反馈, 矩阵分解, 差分隐私保护, 推荐系统

Abstract:

To improve the accuracy of the recommendation algorithm under privacy protection, a logic regression matrix factorization recommendation algorithm is proposed for differential privacy protection. The algorithm first converts the matrix decomposition of implicit data into a classification problem to model it in a probabilistic way. Then, the sigmoid function is used for non-linearly transformation of the prediction score, and the original matrix decomposition problem is converted into two successive user latent factors and item latent factor optimization problem. After that, random noise perturbation is added to the objective function to make the algorithm satisfies differential privacy protection. Experiments are carried out on data sets movielens100k, movielens1m, and Yahoo Music. Compared with the existing relevant algorithms, the algorithm improves the F1 value index by 9.29%, 7.40% and 3.61% respectively. Theoretical analysis and experimental results show that the algorithm can effectively guarantee the accuracy of recommendation results while realizing user implicit feedback data protection, and has good application value.

Key words: implicit feedback, matrix factorization, differential privacy preservation, recommendation system

中图分类号: