Page 314 - 《软件学报》2025年第12期

软件学报 ISSN 1000-9825, CODEN RUXUEW E-mail: jos@iscas.ac.cn
2025,36(12):5695−5719 [doi: 10.13328/j.cnki.jos.007404] [CSTR: 32375.14.jos.007404] http://www.jos.org.cn
©中国科学院软件研究所版权所有. Tel: +86-10-62562563

*
结合时间间隔数据增强的对偶视图自监督会话推荐模型

钱忠胜, 万子珑, 范赋宇, 付庭峰

(江西财经大学 计算机与人工智能学院, 江西 南昌 330013)
通信作者: 钱忠胜, E-mail: changesme@163.com

摘 要: 会话推荐旨在基于用户的一系列项目预测其交互的下一项目, 现有大多数会话推荐对于会话内项目间的
时间间隔信息利用不够充分, 影响推荐准确性. 近年, 图神经网络凭借自身强大的复杂关系建模能力在会话推荐中
受到推崇, 但仅基于图神经网络的会话推荐忽略了会话间的隐藏高阶关系, 信息不够丰富. 此外, 数据稀疏性一直
是推荐系统中存在的现象, 研究中多使用对比学习对此实施改善, 然而大多对比学习框架形式单一, 泛化能力不强.
基于此, 提出一种结合自监督学习的会话推荐模型. 首先, 该模型利用用户会话内项目间的时间间隔信息对会话序
列实施数据增强, 丰富会话内信息, 以提高推荐准确性; 其次, 构建超图卷积网络和 Transformer 编码器相结合的对
偶视图, 从多视角捕捉会话间的隐藏高阶关系, 以丰富推荐多样性; 最后, 融合数据增强后的会话内信息、多视角
下的会话间信息以及原始会话信息进行对比学习, 以增强模型泛化性. 通过与 11 个已有经典模型在 4 个数据集上
的对比发现, 所提模型是可行高效的, 在 HR 与 NDCG 指标上分别平均提升 5.96%、5.89%.
关键词: 会话推荐; 自监督学习; 超图卷积网络; 对偶视图; 数据增强
中图法分类号: TP181

中文引用格式: 钱忠胜, 万子珑, 范赋宇, 付庭峰. 结合时间间隔数据增强的对偶视图自监督会话推荐模型. 软件学报, 2025, 36(12):
5695–5719. http://www.jos.org.cn/1000-9825/7404.htm
英文引用格式: Qian ZS, Wan ZL, Fan FY, Fu TF. Dual-view Self-supervised Session-based Recommendation Model Based on
Temporal Interval Aware Data Augmentation. Ruan Jian Xue Bao/Journal of Software, 2025, 36(12): 5695–5719 (in Chinese). http://
www.jos.org.cn/1000-9825/7404.htm

Dual-view Self-supervised Session-based Recommendation Model Based on Temporal Interval
Aware Data Augmentation
QIAN Zhong-Sheng, WAN Zi-Long, FAN Fu-Yu, FU Ting-Feng
(School of Computer and Artificial Intelligence, Jiangxi University of Finance and Economics, Nanchang 330013, China)
Abstract: Session-based recommendation aims to predict the next item a user will interact with based on a series of items. Most existing
session-based recommender systems do not fully utilize the temporal interval information between items within a session, affecting the
accuracy of recommendations. In recent years, graph neural networks have gained significant attention in session-based recommendation
due to their strong ability to model complex relationships. However, session-based recommendations that rely solely on graph neural
networks overlook the hidden high-order relationships between sessions, resulting in less rich information. In addition, data sparsity has
always been a phenomenon in recommender systems, and contrastive learning is often employed to address this issue. However, most
contrastive learning frameworks lack strong generalization capabilities due to their singular form. Based on this, a session-based
recommendation model combined with self-supervised learning is proposed. First, the model utilizes the temporal interval information
between items within user sessions to perform data augmentation, enriching the information within the sessions to improve
recommendation accuracy. Second, a dual-view encoder is constructed, combining a hypergraph convolutional network encoder and a

* 基金项目: 国家自然科学基金 (62262025); 江西省自然科学基金重点项目 (20224ACB202012); 赣鄱俊才支持计划-主要学科学术和技术
带头人培养项目-领军人才 (学术类) (20243BCE51024)
收稿时间: 2024-08-02; 修改时间: 2024-11-08, 2024-12-14; 采用时间: 2025-01-29; jos 在线出版时间: 2025-06-18
CNKI 网络首发时间: 2025-06-19