 In recent years, knowledge graphs have been used to address cold start and data sparsity issues in recommender systems. This paper proposes Knowledge Aware Enhanced Network, KAN, which combines neighborhood information with knowledge graph to enrich user and item descriptions. The model then aggregates the contributions of different neighbors in the knowledge graph to improve the user and item representations. Additionally, it captures latent distant personalized preferences by propagating them across the knowledge graph. Finally, the model shares potential interaction characteristics between items and entities to further improve recommendations. Experiments on three data sets demonstrate that KAN outperforms state-of-the-art baselines. This article was authored by Xiaolou Wang, Jiu Weiqin, Shang Jiu Deng, and others.