Community detection is a crucial research topic in the realm of complex networks. Understanding and identifying the community structure of a network is essential for uncovering its behavior and function. In this paper, we propose a novel graph em‐ bedding method based on neighbor similarity for community detection. By utilizing the acceptance of nodes and aggregating attri‐ bute information expressions of neighbors, we obtain the vector representation of each node in the network. The final community de‐ tection results are then obtained by directly applying K-means clustering. Our experimental results demonstrate that our proposed al‐ gorithm outperforms other methods, showing significant improvements in both modularity and standard normalization metrics.