Page 205 - 《软件学报》2026年第2期

软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
2026,37(2):684−699 [doi: 10.13328/j.cnki.jos.007419] [CSTR: 32375.14.jos.007419] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563

*
基于多视角的自监督推荐方法

王丹丹 1 , 刘海洋 1 , 刘 壮 2 , 原继东 1

1
(北京交通大学 计算机与信息技术学院, 北京 100044)
2
(北京航空航天大学 计算机学院, 北京 100191)
通信作者: 原继东, E-mail: yuanjd@bjtu.edu.cn

摘 要: 自监督学习可以从原始数据中挖掘自监督信号, 在提高推荐性能方面蕴含着巨大的潜力. 然而, 目前基于
自监督学习的推荐方法存在两个关键的挑战. 首先, 大多数自监督推荐模型采用对同一节点随机扰动的方式, 将生
成的不同结果作为自监督信号, 然而, 由于推荐系统中存在着广泛的同质性, 这种方式会忽略邻居节点信息, 影响
推荐性能. 其次, 用户-物品之间的历史交互信息以及用户与用户之间的社交关系信息是目前基于自监督学习推荐
模型关注的焦点, 而忽略了物品之间的内在联系, 同样会导致产生的自监督信号不够充分. 基于这些挑战, 提出一
种基于多视角的自监督推荐方法, 分别从偏好视角、用户视角、物品视角考虑, 进而使用多视图共同训练的自监
督学习方法, 结合用户之间的社交关系、物品之间的类别关系、用户-物品之间的历史交互信息, 充分挖掘自监督
信号. 在 3 个真实的公开数据集上进行实验, 实验结果验证了基于多视角的自监督学习方法在改进推荐性能方面
是有效的.
关键词: 自监督学习; 推荐系统; 多视角
中图法分类号: TP18

中文引用格式: 王丹丹, 刘海洋, 刘壮, 原继东. 基于多视角的自监督推荐方法. 软件学报, 2026, 37(2): 684–699. http://www.jos.org.
cn/1000-9825/7419.htm
英文引用格式: Wang DD, Liu HY, Liu Z, Yuan JD. Self-supervised Recommendation Method Based on Multiple Views. Ruan Jian
Xue Bao/Journal of Software, 2026, 37(2): 684–699 (in Chinese). http://www.jos.org.cn/1000-9825/7419.htm

Self-supervised Recommendation Method Based on Multiple Views
1
2
1
WANG Dan-Dan , LIU Hai-Yang , LIU Zhuang , YUAN Ji-Dong 1
1
(School of Computer and Information Technology, Beijing Jiaotong University, Beijing 100044, China)
2
(School of Computer Science and Engineering, Beihang University, Beijing 100191, China)
Abstract: Self-supervised learning (SSL) can extract self-supervised signals from raw data, which holds great potential for improving
recommendation performance. However, two key challenges remain in current self-supervised learning-based recommendation methods.
First, most self-supervised recommendation models apply random perturbations to the same node and use the generated different results as
self-supervised signals. However, due to the extensive homogeneity in recommendation systems, this method ignores the information from
neighboring nodes, which affects the recommendation performance. Secondly, while historical interaction information between users and
items as well as the social relationship information between users are the focus of current self-supervised learning-based recommendation
models, the internal relationships between items are often neglected. This also leads to insufficient self-supervised signals. Based on these
challenges, a self-supervised recommendation method based on multiple views is proposed. This method considers perspectives from
preference, user, and item, and employs a multi-view joint training approach for self-supervised learning. By combining the social
relationships between users, the category relationships between items, and the historical interaction information between users and items,
self-supervised signals are fully extracted. Experiments conducted on three real public datasets validate that the proposed multi-view-based

* 基金项目: 中央高校基本科研业务费专项资金 (2023JBZY035)
收稿时间: 2023-03-24; 修改时间: 2024-04-07; 采用时间: 2025-02-26; jos 在线出版时间: 2025-11-13
CNKI 网络首发时间: 2025-11-14