Based on Spectral Clustering Dynamic Community Discovery Algorithm Research in Temporal Network

Abstract

The study of temporal community discovery is an essential research area in social network analysis. As nodes join or leave social networks, the relationships between nodes are establishing or terminating, which affects community structure changes. Given the social networks discovery algorithm of static community lacks the indispensable historical information of network community nodes, resulting in insufficient network structure analysis and clustering information. Based on the community network evolution division events, the paper extracted the priority for analysis and proposed the SC-DCDA: Spectral Clustering Based Temporal Community Discovery Algorithm. According to experimental observation, the SC-DCDA firstly reduced the dimensionality of high-dimensional data leveraging the method of spectral mapping. Secondly, the improved Fuzzy C-means clustering algorithm was adopted to determine the correlation between nodes in temporal social networks and the communities to be discovered, and finally the community structure analysis was performed according to the evolutionary similarity matrix. The ground truth datasets combined with the typically community discovery algorithm metric Modularity Score experimental verification and performance evaluation. The experimental results show that the algorithm metric is well-suited for the temporal datasets, indicating that the proposed algorithm has achieved several better results in information interaction, clustering effect, and accuracy.