Collaborative Filtering Recommender Systems face data sparsity and cold‐start issues, leading to a decrease in their recommendation performance. Therefore, numerous researchers have integrated knowledge graphs and graph convolutional networks into recommender systems to enhance their performance. Through an analysis of previous knowledge graph convolutional network recommendation systems, the following problems have been identified: (1) Some graph convolutional networks only consider the neighborhood aggregation of items while neglecting the neighborhood aggregation of users. (2) User rating differences are not taken into account. To tackle these problems, a Bilateral Attention Mechanism Graph Convolutional Network Recommender Model is proposed. This model effectively captures high‐order neighborhood information for both items and users by utilizing bipartite attention mechanisms on the knowledge graph and interaction graph. Additionally, it incorporates the influence of user ratings on items. Finally, it aggregates item and user information with neighborhood information to predict user ratings for items. The validity of this model was verified through experimental analysis and comparative evaluation of its performance against SPGCN, CACF on the MovieLens‐20M, Book‐Crossing, and Last.FM datasets.