High-performance social networking: microblog community detection based on efficient interactive characteristic clustering

With the development of microblog networks and the popularity of the friend circle, more and more users are linked together to form communities. The microblog community detection is not only the separation of following relationships. The interactive characteristic between users should also be considered. Therefore, in the article the maximum likelihood estimation is used to extract the interactive characteristics clustering. The Link Optimization Comm (LOC) algorithm based on density clustering is proposed to improve the performance the Link Comm (LC) algorithm. By integrating the interactive characteristic clustering into LOC algorithm, the high-performance IC-LOC method to detect potential communities is proposed. Then the complexity in clustering is analyzed. Simulation experiments show that LOC algorithm is better than LC algorithm in time complexity and normalized mutual information evaluation. Compared with LC algorithm in Sina microblog data sets, the IC-LOC method also achieves better performance of community detection. Moreover, the proposed method can effectively detect interactive and potential communities.