Recently, the “K-Graph-Centers” algorithm proposed by Dr. Chen Wei's graph computing team starts from the basic concept of a "Virtual Graph Center" and, through rigorous mathematical derivation, ultimately evolves into a two-step iterative algorithm almost completely symmetric to k-means. On graph clustering tasks, it achieves slightly better accuracy than classical spectral clustering (Ncut/Rcut) while reducing time cost by two to three orders of magnitude. On an ordinary laptop, for a 12,000 × 12,000 graph, it converges in only about 1 second. This work has been accepted by the IEEE TKDE journal.