| 引用本文: | 汪玉洁1 ,刘 涛1 ,包象琳1 ,潘正高.基于社区划分的社交推荐隐私保护方法(J/M/D/N,J:杂志，M：书，D：论文，N：报纸).期刊名称,2024，41（6）：30-38 |
| CHEN X. Adap tive slidingmode contr ol for discrete2ti me multi2inputmulti2 out put systems[ J ]. Aut omatica, 2006, 42(6): 4272-435 |
|
| 摘要: |
| 目的 社交推荐是在传统推荐的基础上引入用户的社交信息以更好地生成推荐结果。 由于社交推荐不仅涉
及用户本身的信息,还涉及用户的社交关系信息,因此对用户的隐私保护变得更加重要。 然而,目前的社交推荐方
法大多只注重提高推荐准确性,而忽视了对用户个人信息隐私保护的问题。 因此针对社交推荐中用户的评分数据
和社交关系数据的隐私保护问题,提出了一种基于社区划分的社交推荐隐私保护方法( SRCD) 。 方法 首先,考虑
评分值的分布范围对用户相似度的影响,并结合用户之间的社交关系,来给社交网络中的用户划分社区,并计算每
个社区中用户对所看过项目的评分的均值;然后,根据社区划分的结果,寻找与目标用户所在社区相似的其他社
区。 从而可以构造出一个社区-项目评分均值矩阵。 并且针对实际场景中评分均值矩阵稀疏的情况,采用了中位
数填补矩阵的缺失元素。 最后,用矩阵分解的结果来预测用户对项目的评分,从而评估算法的性能。 结果 通过仿
真实验验证,所提方法相比于现有的社交推荐算法不仅在隐私保护方面提供了保障,而且在推荐准确度方面具有
相近的预测准确率。 结论 提出的方法不仅在一定程度上保护了用户的隐私信息,还为用户提供了令人满意的推荐
结果。 |
| 关键词: 隐私保护 社交推荐 社区划分 分组聚合 矩阵分解 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| A Privacy Protection Method for Social Recommendation Based on Community Division |
|
WANG Yujie1 LIU Tao1 BAO Xianglin1 PAN Zhenggao
|
|
1. School of Computer and Information Anhui University of Engineering Anhui Wuhu 241000 China
2. School of Information Engineering Suzhou University Anhui Suzhou 234000 Chin
|
| Abstract: |
| Social recommendation introduces users?? social information based on traditional recommendation to
generate better recommendation results. As social recommendation involves not only the user?? s information but also the
user?? s social relationship information privacy protection for users becomes more important. However most current social
recommendation methods focus only on improving recommendation accuracy and overlook the issue of privacy protection for
users?? personal information. Therefore a privacy protection method for the social recommendation based on community
division SRCD was proposed to address the privacy protection issues of users?? rating data and social relationship data in
the social recommendation. Methods First considering the impact of rating value distribution on user similarity and
combined with the social relationships between users communities in the social network were divided and the average
ratings of users for items they have viewed in each community were calculated. Then based on the results of community division similar communities to the community where the target users belong were identified. This allowed for the
construction of a community-item rating average matrix. Additionally for sparse rating average matrices in practical
scenarios median imputation was used to fill in missing elements in the matrix. Finally matrix decomposition results were
used to predict user ratings for items thus evaluating the algorithm ?? s performance. Results Through simulation
experiments it was verified that the proposed method not only provides a guarantee for privacy protection but also has a
similar prediction accuracy in recommendation accuracy compared with existing social recommendation algorithms.
Conclusion The proposed method not only protects users ?? privacy information to a certain extent but also provides
satisfactory recommendation results for users. |
| Key words: privacy protection social recommendation community division grouping aggregation matrix decomposition |
