Research On Graph Neural Network Recommender Based On Social Relationship And Attention Mechanism

As an important tool to assist users in decision-making,recommender system can extract the information that users may be interested in from the mass of information through a specific algorithm.However,the traditional recommendation model has some problems,such as incomplete use of information,poor recommendation performance and so on.Although the recommendation model based on deep learning is more effective,it can not deal with non-European data.Based on the analysis and summary of traditional recommendation methods,this paper proposes a graph neural network recommendation model which integrates social relationship and attention mechanism by combining theoretical analysis,numerical simulation and experimental methods and combining with the existing graph neural network model.The main contents of this paper are as follows(1)In order to solve the problem that the traditional Variational Graph Autoencoder(VGAE)does not distinguish the neighborhood nodes in the coding process,this paper proposes an improved graph neural network model,which is called AVGAE(Attention Variational Graph Autoencoder)by using multi attention mechanism,Experiments show that it can effectively identify the importance of different entity objects to the target node,which is more suitable for the actual recommendation application scenarios.(2)In view of the problem that the existing recommendation methods do not fully consider the user's social relationship information,this paper proposes a graph variational self encoder recommendation algorithm(Social-AVGAE)based on AVGAE by constructing the social relationship graph neural network model and the user article graph neural network at the same time,and using different mapping relations.Experiments on relevant public data sets show that the graph neural network,which integrates social relations and attention mechanism,has better performance in recommendation performance,and has a certain practical significance in practical application.