With the development of data processing technology, complex network theory has been widely applied in many areas. Meanwhile, as one of the essential parts of network science, community detection is becoming more and more important for analyzing and visualizing the real world. Specially, signed network is a kind of graph which can more truly and efficiently reflect the reality, however, the study of community detection on signed network is still rare. In this paper, we propose a new agglomerative algorithm based on the modularity optimization for community detection on signed networks. The proposed model utilizes a new data structure called community adjacency list in signed (CALS) networks to improve the efficiency. Successive modularity computations make the connections between node changes so that the process time leads to substantial savings. Experiments on both real and artificial networks verify the accuracy and efficiency of this method, which is suitable for the application on large-scale networks.
A Neighborhood-Impact Based Community Detection Algorithm via Discrete PSO
