Recommendation Algorithm Based on Social Information and Knowledge Graph Attention Network
XU Changlin
WANG Xun
HUANG Shucheng
Abstract：In order to solve the problem that the recommendation algorithm based on knowledge graph ignores the correlation between users when using semantic information,resulting in the lack of neighborhood user information in the user's feature expres-sion,a recommendation algorithm(SKGAN)based on social information and knowledge graph attention network is proposed.Through knowledge embedding,the low-dimensional representation of the entity is obtained in the algorithm.The user features are obtained by performing convolution of the user-item bipartite graph,propagated and aggregated through social networks.An atten-tion mechanism is used to calculate the weights in the process of propagation.The enhanced user expression is combined with proj-ect representation of convolutional output of the item knowledge graph,so as to achieve recommendations.The experimental results show that the SKGAN algorithm outperforms the baseline model in both AUC and F1-Score metrics.
Keywords：recommender systemsocial networkknowledge graphgraph convolutional networkattention mechanism
Publication Date：2025-05-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 1215-1221 )