Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (22): 111-120.DOI: 10.3778/j.issn.1002-8331.2207-0314

• Pattern Recognition and Artificial Intelligence • Previous Articles     Next Articles

Neural Collaborative Recommendation Algorithm Based on Attention Mechanism and Knowledge Graph

ZHANG Chuang, WANG Wei, DU Yuxuan, ZHENG Xiaoli, HE Tingting   

  1. 1.School of Information & Electrical Engineering, Hebei University of Engineering, Handan, Hebei 056038, China
    2.Hebei Key Laboratory of Security & Protection Information Sensing & Processing, Hebei University of Engineering, Handan, Hebei 056038, China
    3.School of Internet of Things Engineering, Jiangnan University, Wuxi, Jiangsu 214122, China
  • Online:2023-11-15 Published:2023-11-15



  1. 1.河北工程大学 信息与电气工程学院,河北 邯郸 056038
    2.河北工程大学 河北省安防信息感知与处理重点实验室,河北 邯郸 056038
    3.江南大学 物联网工程学院,江苏 无锡 214122

Abstract: At present, the traditional recommendation algorithm based on collaborative filtering has poor performance in the face of sparse data and cold start. However, the recommendation system assisted by knowledge graph can effectively alleviate this problem. Supplemented by attention mechanism, a neural collaborative recommendation algorithm combining attention mechanism and knowledge graph is designed. Then the attention mechanism is used to learn and aggregate the higher-order potential relationship information in the knowledge graph. At the same time, the user’s final preference is obtained through gated recurrent neural network training based on the user’s long-term and short-term interest preferences. Finally, the collaborative filtering method is used to generate the recommendation list. Through experiments on MovieLens-1M and Amazon-Book datasets, the recommended recall rate, accuracy rate, hit rate and NDCG evaluation indicators are all improved, which validates the effectiveness of the proposed method.

Key words: knowledge graph, attention mechanism, neural network, collaborative filtering

摘要: 目前传统的基于协同过滤的推荐算法面对数据稀疏和冷启动问题时表现欠佳,而知识图谱辅助的推荐系统可以有效缓解这一问题,辅以注意力机制,设计了一种融合注意力机制与知识图神经协同推荐算法。首先根据用户与项目的交互图与项目属性的知识图谱进行组合,以此为基础进行嵌入表示;然后利用注意力机制来学习知识图谱中的高阶潜在关系信息并进行聚合,同时结合用户的长短期兴趣偏好通过门控循环神经网络训练获取用户最终偏好进行推荐;最后采用协同过滤方法生成推荐列表。在MovieLens-1M和Amazon-Book数据集上进行实验,所提算法在推荐召回率、准确率、命中率和NDCG的评价指标上均有提升,验证了算法的有效性。

关键词: 知识图谱, 注意力机制, 神经网络, 协同过滤