The explosive growth of Internet film and television resources has caused users to fall into information overload and selection dilemma. However, the traditional recommendation system is difficult to meet the actual needs due to data sparsity, cold start and insufficient interpretability. Although knowledge graph and graph neural network technology provide a new path for recommendation system optimization, the existing research still has deficiencies in the quality of graph construction, model selection and coordination mechanism, especially the lack of systematic comparative analysis of mainstream graph neural network models. To this end, this paper constructs a knowledge graph that integrates user, movie and type information, and compares graph convolutional network, GraphSAGE, graph attention network and Heterogeneous Graph Neural Network (HGNN) on three MovieLens datasets of different sizes. The experimental results show that HGNN exhibits optimal performance in both rating prediction and Top-K recommendation tasks. Its mean square error and mean absolute error on the ml-10 m dataset are as low as 0.861 3 and 0.719 7, respectively. Precision@5 and NDCG@5 are 0.335 1 and 0.429 5, respectively. The research results confirm the advantages of HGNN in modeling heterogeneous semantic relations, and provide a key empirical basis for the model selection of knowledge graph enhancement recommendation system.
Top-K推荐是指从海量的电影中,为用户推荐最可能喜欢的前K部电影,其中,K表示一个正整数(如K=5,10,15等)。例如,Top-5推荐:为用户推荐最可能喜欢的前5部电影。Top-K推荐任务中采用精确度Precision@K和归一化折损累计增益NDCG@K(Normalized Discounted Cumulative Gain at K)作为评价指标。
WANGZ, WANGZ L, LIX,et al. Exploring multi-dimension User-item interactions with attentional knowledge graph neural networks for recommendation[J]. IEEE Transactions on Big Data,2022,9(1):212-226.
[4]
BHATTIU A, TANGH, WUG,et al. Deep learning with Graph Convolutional Networks:An overview and latest applications in computational intelligence[J]. International Journal of Intelligent Systems,2023,2023:8342104.
[5]
侯映鑫. 基于图神经网络的电影推荐系统设计与实现[D]. 兰州:西北民族大学,2022.
[6]
KIPFT N, WELLINGM. Semi-Supervised classification with graph convolutional networks[C]//Proceedings of the International Conference on Learning Representations (ICLR). Toulon:ICLR,2017:1-9.
[7]
HAMILTONW L, YINGR, LESKOVECJ. Inductive representation learning on large graphs[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS 2017). Long Beach:Curran Associates Incorporated,2017:1025-1035.
[8]
VELIČKOVIĆP, CUCURULLG, CASANOVAA,et al. Graph attention networks[R]. Ithaca:Cornell University Library,2017.
[9]
ZHANGJ, SHIX, XIEJ,et al. Heterogeneous graph neural network[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. Anchorage:ACM,2019:793-803.
[10]
BASTOSA, NADGERIA, SINGHK,et al. RECON:Relation extraction using knowledge graph context in a graph neural network[C]//Proceedings of the Web Conference. Ljubljana:ACM,2021:1673-1685.
[11]
YER, LIX, FANGY,et al. A vectorized relational graph convolutional network for multi-relational network alignment[C]//2019 IEEE International Conference on Data Mining (ICDM). Beijing:IEEE,2019:4135-4141.
[12]
刘映宏. 基于知识图谱与图神经网络的电影推荐系统设计与实现[D]. 成都:电子科技大学,2025.
[13]
牛妍辉. 基于知识图谱的个性化电影推荐系统的研究与实现[D]. 石河子:石河子大学,2023.
[14]
刘英东. 基于知识图谱的电影推荐系统研究[D]. 太原:山西大学,2023.
[15]
MISHRAR, SHRIDEVIS. Knowledge graph driven medicine recommendation system using graph neural networks on longitudinal medical records[J]. Scientific Reports,2024,14(1):25449.
[16]
NGUYEND A,KHA S, LET V. HybridGCN:An integrative model for scalable recommender systems with knowledge graph and graph neural networks[J]. International Journal of Advanced Computer Science and Applications,2024,15(5):1327-1337.
NESMAOUIR, LOUHICHIM, LAZAARM. A collaborative filtering movies recommendation system based on graph neural network[J]. Procedia Computer Science,2023,220:456-461.
[20]
LIW, ZHONGH, ZHOUJ,et al. An attention mechanism and residual network based knowledge graph-enhanced recommender system[J]. Knowledge-Based Systems,2024,299:112042.
[21]
刘松. 基于异质图神经网络的电影推荐系统研究[D]. 重庆:重庆邮电大学,2022.
[22]
SUNR, ZHANGW, MAY,et al. MKGAT:Multi-modal knowledge graph attention network for recommendation[C]//Proceedings of the 29th ACM International Conference on Information and Knowledge Management (CIKM 2020). New York:ACM,2020:1483-1492.