A Community Detecting Algorithm Based on Modular Tensor in Temporal Network.

With the emergence of high-dimensional data and a deeper research in complex network, scholars began to analyze the features of multi-dimensional network, such as the high-dimension community structure. Especially in the temporal network, many researchers have already proposed community discovery algorithms based on tensor decomposition. However, there are still some deficiencies in current research, which is the disconnection between traditional community detecting theory and the algorithm based on tensor decomposition. To solve this problem, this paper proof that there is a relationship between the modularity in the community discovery and the optimization goal of tensor decomposition. Based on the proof, a new community discovery algorithm based on modularity tensor in temporal network is proposed. Experiments in simulated dataset and real PSTN dataset have proved that our proposed algorithm can find a better community division than the traditional algorithm.