先基于系统网络扩散，生成每个图中节点对网络的影响，再算相似度，最后使用k均值聚类-First diffusion-based system network, generated for each graph node on the network effect, and then calculate the similarity, and finally use the k-means clustering