Community structure detection based on the neighbor node degree information

Community structure detection is of great significance for better understanding the network topology property. By taking into account the neighbor degree information of the topological network as the link weight, we present an improved Nonnegative Matrix Factorization (NMF) method for detecting community structure. The results for empirical networks show that the largest improved ratio of the Normalized Mutual Information value could reach [Formula: see text]. Meanwhile, for synthetic networks, the highest Normalized Mutual Information value could closely reach 1, which suggests that the improved method with the optimal [Formula: see text] can detect the community structure more accurately. This work is helpful for understanding the interplay between the link weight and the community structure detection