Abstract: According to the problems of community-finding on complex networks, this paper detailed studies the clustering method to it, changes the data structure of the network node into that of clustering algorithm, amalgamating or partitioning nodes according to the similarity of nodes, and uses the vector method to measure the similarity of complex network nodes. The improved algorithm makes each node on network as a source having sending and receiving messages function, transfers and traverses similarity value according to the improved information transmission method, uses Karate club network commonly uses on complex network as verification of the experimental object. It improves the efficiency of community-finding on complex networks.

Key words: community-finding, clustering algorithm, similarity between nodes, node quantifying