Graph convolutional recommender system integrating contrastive learning and global graph data augmentation

doi:10.11772/j.issn.1001-9081.2025091126

Abstract

Abstract:

With the rapid growth of information， recommender systems have played a crucial role in improving user experience and solving information overload. However， the traditional recommendation methods still face limitations in addressing data sparsity， cold-start problems， and model interpretability. To tackle these challenges， a graph contrastive learning convolutional neural network recommender system based on global graph data augmentation， called GACL-Rec （Global Augmented Contrastive Learning for Recommendation）， was proposed. First， global augmentation was performed at the node， edge， and feature levels to enhance the model’s robustness under sparse data conditions. Second， a multi-granularity graph contrastive learning framework was designed， in which the consistency and discriminability of the embeddings were strengthened through contrastive learning tasks at the node， subgraph， and graph levels. Experimental results on several public datasets， including MovieLens-1M and Amazon-Books， demonstrate that the proposed recommender system outperforms the traditional models such as Matrix Factorization （MF） and LightGCN in terms of recommendation accuracy， robustness， and performance in cold-start scenarios， particularly achieving superior results on sparse datasets. Furthermore， parameter sensitivity analyses and ablation study results verify the effectiveness of the global augmentation strategy and the multi-granularity contrastive learning mechanism. It can be seen that the proposed model has significant advantages in addressing key challenges in recommender systems， such as data sparsity and cold-start problems.

Key words: recommender system, Graph Neural Network (GNN), data augmentation, contrastive learning, cold-start problem

摘要：

随着信息量的迅速增长，推荐系统在改善用户体验和解决信息过载方面发挥了重要作用。然而，传统推荐方法在数据稀疏性、冷启动问题和模型可解释性方面仍存在局限。为了解决这些问题，提出一种基于全局图数据增强的图对比学习卷积神经网络推荐系统GACL-Rec（Global Augmented Contrastive Learning for Recommendation）。首先，在节点、边和特征层面进行全局增强，以提升模型在稀疏数据下的鲁棒性；其次，设计多粒度图对比学习框架，通过节点级、子图级和全图级的对比学习任务，增强嵌入表示的一致性与判别性。在MovieLens-1M、Amazon-Books等多个公开数据集上的实验结果表明，所提推荐系统在推荐精度、鲁棒性和冷启动场景中的表现优于矩阵分解（MF）与LightGCN等传统模型，在稀疏数据集上的表现尤为突出。此外，通过参数敏感性分析和消融实验结果验证了全局增强和多粒度对比学习机制的有效性。可见，所提模型在解决推荐系统中的典型问题（如数据稀疏和冷启动问题）时具有显著优势。

关键词: 推荐系统, 图神经网络, 数据增强, 对比学习, 冷启动问题

CLC Number:

TP311.13

Long CHEN, Xinmin HOU. Graph convolutional recommender system integrating contrastive learning and global graph data augmentation[J]. Journal of Computer Applications, 2026, 46(10): 3076-3082.

陈龙, 侯新民. 基于全局图数据增强的对比学习图卷积推荐系统[J]. 《计算机应用》唯一官方网站, 2026, 46(10): 3076-3082.

Figures/Tables 10

References 30

[1]	Goldberg D， Nichols D， Oki B M， et al. Using collaborative filtering to weave an information tapestry［J］. Communications of the ACM， 1992， 35（12）： 61-70.
[2]	Koren Y， Bell R， Volinsky C. Matrix factorization techniques for recommender systems［J］. Computer， 2009， 42（8）： 30-37.
[3]	He X， Deng K， Wang X， et al. LightGCN： simplifying and powering graph convolution network for recommendation［C］// SIGIR 2020. New York： ACM， 2020： 639-648.
[4]	Wang X， He X， Wang M， et al. Neural graph collaborative filtering［C］// SIGIR 2019. New York： ACM， 2019： 165-174.
[5]	Li Q， Han Z， Wu X M. Deeper insights into graph convolutional networks for semi-supervised learning［J］. Proceedings of the AAAI Conference on Artificial Intelligence， 2018， 32（1）： 3538-3545.
[6]	Qiu R， Li J， Huang Z， et al. Rethinking the item order in session-based recommendation with graph neural networks［C］// CIKM 2019. New York： ACM， 2019： 579-588.
[7]	Wu S， Tang Y， Zhu Y， et al. Session-based recommendation with graph neural networks［J］. Proceedings of the AAAI Conference on Artificial Intelligence， 2019， 33（1）： 346-353.
[8]	Chang B， Jang G， Kim S， et al. Learning graph-based geographical latent representation for point-of-interest recommendation［C］// CIKM 2020. New York： ACM， 2020： 135-144.
[9]	Lim N， Hooi B， Ng S K， et al. STP-UDGAT： spatial-temporal-preference user dimensional graph attention network for next POI recommendation［C］// CIKM 2020. New York： ACM， 2020： 845-854.
[10]	He Z， Chow C Y， Zhang J D. GAME： learning graphical and attentive multi-view embeddings for occasional group recommendation［C］// SIGIR 2020. New York： ACM， 2020： 649-658.
[11]	Wang W， Zhang W， Rao J， et al. Group-aware long- and short-term graph representation learning for sequential group recommendation［C］// SIGIR 2020. New York： ACM， 2020： 1449-1458.
[12]	Chang J， Gao C， He X， et al. Bundle recommendation with graph convolutional networks［C］// SIGIR 2020. New York： ACM， 2020： 1673-1676.
[13]	Eksombatchai C， Jindal P， Liu J Z， et al. Pixie： a system for recommending 3+ billion items to 200+ million users in real-time［C］// WWW 2018. Republic and Canton of Geneva： International World Wide Web Conferences Steering Committee， 2018： 1775-1784.
[14]	Pfadler A， Zhao H， Wang J， et al. Billion-scale recommendation with heterogeneous side information at Taobao［C］// ICDE 2020. Piscataway： IEEE， 2020： 1667-1676.
[15]	Ding K， Wang J， Li J， et al. Graph prototypical networks for few-shot learning on attributed networks［C］// CIKM 2020. New York： ACM， 2020： 295-304.
[16]	Zhao T， Liu Y， Neves L， et al. Data augmentation for graph neural networks［J］. Proceedings of the AAAI Conference on Artificial Intelligence， 2021， 35（12）： 11015-11023.
[17]	Shorten C， Khoshgoftaar T M. A survey on image data augmentation for deep learning［J］. Journal of Big Data， 2019， 6： No.60.
[18]	Ding K， Wang J， Caverlee J， et al. Meta propagation networks for graph few-shot semi-supervised learning［J］. Proceedings of the AAAI Conference on Artificial Intelligence， 2022， 36（6）： 6524-6531.
[19]	Liu X， Zhang F， Hou Z， et al. Self-supervised learning： generative or contrastive［J］. IEEE Transactions on Knowledge and Data Engineering， 2023， 35（1）： 857-876.
[20]	Kipf T N， Welling M. Semi-supervised classification with graph convolutional networks［PP/OL］. V4. arXiv （2017-02-22）［2025-07-20］..
[21]	Hamilton W L， Ying R， Leskovec J. Inductive representation learning on large graphs［C］// NeurIPS 2017. Red Hook： Curran Associates Inc.， 2017： 1025-1035.
[22]	You Y， Chen T， Sui Y， et al. Graph contrastive learning with augmentations［C］// NeurIPS 2020. Red Hook： Curran Associates Inc.， 2020： 5812-5823.
[23]	van den Oord A， Li Y， Vinyals O. Representation learning with contrastive predictive coding［PP/OL］. V2. arXiv （2019-01-22）［2025-07-20］..
[24]	Chen T， Kornblith S， Norouzi M， et al. A simple framework for contrastive learning of visual representations［J］. Proceedings of Machine Learning Research， 2020， 119： 1597-1607.
[25]	Zhu Y， Xu Y， Yu F， et al. Graph contrastive learning with adaptive augmentation［C］// WWW 2021. New York： ACM， 2021： 2069-2080.
[26]	Yu J， Yin H， Xia X， et al. Are graph augmentations necessary？ simple graph contrastive learning for recommendation［C］// SIGIR 2022. New York： ACM， 2022： 1294-1303.
[27]	Zhou C， Ma J， Zhang J， et al. Contrastive learning for debiased candidate generation in large-scale recommender systems［C］// SIGKDD 2021. New York： ACM， 2021： 3985-3995.
[28]	Harper F M， Konstan J A. The MovieLens datasets： history and context［J］. ACM Transactions on Interactive Intelligent Systems， 2015， 5（4）： No.19.
[29]	Ni J， Li J， McAuley J. Justifying recommendations using distantly-labeled reviews and fine-grained aspects［C］// EMNLP-IJCNLP 2019. Stroudsburg： ACL， 2019： 188-197.
[30]	Zhang X， Zhao J， LeCun Y. Character-level convolutional networks for text classification［C］// NeurIPS 2015. Cambridge： MIT Press， 2015： 649-657.

数据集	用户数	物品数	交互数	稀疏率/%
MovieLens-1M	6 040	3 952	1 000 209	95.80
MovieLens-100K	943	1 682	100 000	93.70
Amazon-Books	52 406	41 264	8 898 041	99.60
Yelp	45 981	11 537	1 185 068	99.77

数据集	用户数	物品数	交互数	稀疏率/%
MovieLens-1M	6 040	3 952	1 000 209	95.80
MovieLens-100K	943	1 682	100 000	93.70
Amazon-Books	52 406	41 264	8 898 041	99.60
Yelp	45 981	11 537	1 185 068	99.77

模型	MovieLens-1M		Amazon-Books		Yelp
模型	Recall@20	NDCG@20	Recall@20	NDCG@20	Recall@20	NDCG@20
GCN	0.312	0.186	0.068	0.029	0.051	0.021
NGCF	0.346	0.211	0.079	0.037	0.064	0.028
LightGCN	0.374	0.232	0.092	0.042	0.073	0.032
GraphCL	0.382	0.239	0.099	0.048	0.079	0.037
SimGCL	0.389	0.245	0.103	0.051	0.083	0.039
GACL-Rec	0.401	0.257	0.117	0.059	0.091	0.044

模型	MovieLens-1M		Amazon-Books		Yelp
模型	Recall@20	NDCG@20	Recall@20	NDCG@20	Recall@20	NDCG@20
GCN	0.312	0.186	0.068	0.029	0.051	0.021
NGCF	0.346	0.211	0.079	0.037	0.064	0.028
LightGCN	0.374	0.232	0.092	0.042	0.073	0.032
GraphCL	0.382	0.239	0.099	0.048	0.079	0.037
SimGCL	0.389	0.245	0.103	0.051	0.083	0.039
GACL-Rec	0.401	0.257	0.117	0.059	0.091	0.044

模型设置	Recall@20	NDCG@20
完整模型（GACL-Rec）	0.612	0.355
去除全局增强（w/o Global Aug.）	0.563	0.329
去除节点级对比学习（w/o Node-CL）	0.570	0.334
去除子图级对比学习（w/o Subgraph-CL）	0.582	0.341
去除全图级对比学习（w/o Global-CL）	0.589	0.345