Journal of Computer Applications ›› 2023, Vol. 43 ›› Issue (9): 2741-2746.DOI: 10.11772/j.issn.1001-9081.2022091361

• Artificial intelligence • Previous Articles     Next Articles

Collaborative recommendation algorithm based on deep graph neural network

Runchao PAN, Qishan YU, Hongfei XIONG, Zhihui LIU()   

  1. School of Mathematics and Physics,China University of Geoscience,Wuhan Hubei 430074,China
  • Received:2022-09-12 Revised:2022-11-10 Accepted:2022-11-14 Online:2023-02-14 Published:2023-09-10
  • Contact: Zhihui LIU
  • About author:PAN Runchao, born in 1997, M. S. candidate. His research interests include recommendation system.
    YU Qishan, born in 1997, M. S. candidate. His research interests include recommendation system.
    XIONG Hongfei, born in 1997, M. S. candidate. His research interests include recommendation system.
  • Supported by:
    National Natural Science Foundation of China(12001408);Science and Technology Research Project of Department of Education of Hubei Province(B2018541)


潘润超, 虞启山, 熊泓霏, 刘智慧()   

  1. 中国地质大学(武汉) 数学与物理学院,武汉 430074
  • 通讯作者: 刘智慧
  • 作者简介:潘润超(1997—),男,江苏常州人,硕士研究生,主要研究方向:推荐系统
  • 基金资助:


For the problem of over-smoothing in the existing recommendation algorithms based on Graph Neural Network (GNN), a collaborative filtering recommendation algorithm based on deep GCN was proposed, namely Deep NGCF (Deep Neural Graph Collaborative Filtering). In the algorithm, the initial residual connection and identity mapping were introduced into GNN, which avoided GNN from falling into over-smoothing after multiple graph convolution operations. Firstly, the initial embeddings of users and items were obtained through their interaction history. Next, in aggregation and propagation layer, collaborative signals of users and items in different stages were obtained with the use of initial residual connection and identity mapping. Finally, score prediction was performed according to the linear representation of all collaborative signals. In addition, to further improve the flexibility and recommendation performance of the model, the weights were set in the initial residual connection and identity mapping for adjustment. In order to verify the feasibility and effectiveness of Deep NGCF algorithm, experiments were conducted on datasets Gowalla, Yelp-2018 and Amazon-book. The results show that compared with the existing GNN recommendation algorithm such as Graph Convolutional Matrix Completion (GCMC) and Neural Graph Collaborate Filtering (NGCF), Deep NGCF algorithm achieves the best results on recall and Normalized Discounted Cumulative Gain (NDCG), thereby verifying the effectiveness of the proposed algorithm.

Key words: recommendation algorithm, collaborative filtering, Graph Neural Network (GNN), identity mapping, initial residual connection


针对现有基于图神经网络(GNN)的推荐算法面临的过平滑的问题,提出一种基于深度GNN的协同过滤推荐算法Deep NGCF(Deep Neural Graph Collaborative Filtering)。该算法在GNN中引入初始残差连接和恒等映射,避免了GNN进行多次图卷积运算后陷入过平滑。首先,通过用户和项目的交互历史得到它们的初始嵌入;其次,在聚合传播层利用初始残差连接和恒等映射得到用户和项目的不同阶协同信号;最后,对所有协同信号进行线性表示以得到预测评分。此外,在初始残差连接和恒等映射中设置比重进行调节,从而进一步提高模型的灵活性和推荐性能。为验证Deep NGCF算法的可行性和有效性,在Gowalla、Yelp-2018与Amazon-Book数据集上进行实验。实验结果表明,相较于图卷积矩阵补全(GCMC)、神经图协同过滤(NGCF)等现有的GNN推荐算法,Deep NGCF算法取得了最高的召回率和归一化折损累计增益(NDCG),验证了所提算法的有效性。

关键词: 推荐算法, 协同过滤, 图神经网络, 恒等映射, 初始残差连接

CLC Number: