Fast Partitioning Algorithm for Detecting Communities in Complex Networks

Detecting Communities is an important research field for both theoretical research and practical application in complex networks.Most of the proposed splitting algorithms and spectrum division algorithms are not suitable for very large networks because of their high time complexity and unknown quantity of community number.We propose a fast partitioning algorithm based on diffusion distance and the modularity function.Its Spectrum division basis is the diffusion distance,and the ability of modularity function can find the best community number in large networks.Furthermore,the convergence of the diffusion distance is also tested.Experimental results show that our algorithm has better partitioning ability and lower time complexity than the existing partitioning community structure algorithms.Not only is it capable of conducting fast operations for the sparse network,but also for the non-sparse network,which reflects the algorithm has high robustness.