Community Detection Algorithm for Dynamic Academic Network

• Published in: Ji suan ji ke xue Vol. 49; no. 1; pp. 89 - 94
• Main Authors: Pu, Shi, Zhao, Wei-dong
• Format: Journal Article
• Language: Chinese
• Published: Chongqing Guojia Kexue Jishu Bu 01.01.2022; Editorial office of Computer Science
• Subjects: academic network|dynamic network|community detection|heterogeneous network|clustering optimization; Algorithms; Clustering; Dynamic characteristics; Faults; Machine learning; Modularity; Optimization; Representations
• ISSN: 1002-137X
• DOI: 10.11896/jsjkx.210100023

The scientific research network is a kind of dynamically changing heterogeneous information network. Community detection on the scientific research network can dig out the community of academic subjects and discover the insights contained in the scientific research community. The existing community detection algorithms ignore the dynamic characteristics of the scientific research network. Due to the special relationship with scientific research subjects, the closeness within the scientific research community and the relationship between communities are not integrated into the community detection algorithm for optimization. To this end, a community detection algorithm DANE-CD based on dynamic scientific research network representation learning is proposed. Based on the scientific research network autoencoder to learn the representation vector of academic subjects in the scientific research network, and then innovatively integrates the clustering optimization based on the two dimensions of modularity and team f