Abstract: Since the existing community analysis methods cannot be applied in large-scale networks， a community extraction algorithm is proposed. The community structure can be analyzed effectively with the algorithm. The topology of the network is not needed， with the combination of network search and community detection capabilities， the structure of the particular community can be effectively extracted from the social network with unknown topology. The analyzing of the community structure of large scale network is possible with the algorithm. Experiments on simulation data are performed to analyze the influence factor of accuracy， and it is concluded that the accuracy increases with the average degree. Furthermore， it is found that the accuracy of community extraction algorithm is close to existing methods， and the efficiency is much better， the results show the algorithm is feasible and effective.

Key words: social network, community extraction, community detection, community structure, network search