Research In Complex Network Community Detection And Critical Node Recognition For Biological Fields

In recent years,an increasing literature on the biology has been available,which involves in broad research fields and numerous hot pots.The publications in terms of biology tend to combine various disciplines and then propose higher requirements on research.Based on the above backgrounds,the study of this paper applies dynamic community detection aiming at complex networks,based on the biology heterogeneous networks of publications from PubMed,to analyze the citation relationship among authors and papers,the development of the research groups,and so on.Concretely,the main work is as follows:Firstly,we construct the complex multi-layer heterogeneous networks of biology research,which contain paper layer,author layer and affiliation layer.We introduce and give the formal descriptions of the each layer’s network structure.And we also introduce the formal definitions of characteristics of complex networks and analyze the properties of the author networks and the research group networks.Next,we propose the dynamic community detection algorithm PPNMF based on nonnegative matrix factorization,and analyze the biosafety networks using the method we proposed.To recognize critical nodes in detected communities,we apply several evaluation metrics in bibliometrics.We present the results of community detection on biology multi-layer heterogeneous complex networks by applying visualization tools.We identify the key nodes of researchers and groups and present the results of the dynamic community detection algorithm.In conclusion,we construct the biosafety multi-layer heterogeneous complex networks and propose the dynamic community detection on complex networks.In addition,we analyze the statistical characteristics of researchers and research groups,and discover their community structures.We identify the critical nodes in detected communities and analyze the relationships among them.tion tools.We identify the key nodes of networks and present the results of the community detection algorithm.