Community division is an important research topic in complex network area. In order to quickly and accurately find community structures in complex network, a similarity based community division algorithm named SCDA is proposed in this paper. This algorithm finds community structures by clustering the nodes according to the importance of nodes and their similarities. It selects the node owning greater clustering coefficient as the cluster center, puts the nodes with similarity larger than given threshold to the cluster, and iterates the process until the node collection is empty. Then the clusters generated by the algorithm are communities. SCDA reduces the time complexity by starting from the important node and ignoring those nodes having been clustered when determing new cluster center, and it promotes the division accuracy by using clustering coefficient and node similarity properly. The experimental result of applying the algorithm to the classical social network, the Zachary's Network of Karate Club Members, shows that SCDA costs less time and has higer accuracy. The algorithm SCDA is valid in community division.
展开▼
