Complex Co-authorship Network Analysis And Visualization

With the rapid development of scientific research cooperation,almost all of the academic articles are published by several authors co-authored.As one of the most representative complex networks,co-authorship network is often processed by social network analysis method,and this method has become today's main technical means.Based on co-authorship network data,evaluation of main problems in research of co-authorship network behavior analysis and visualization,research in this thesis focuses on the study of co-authorship network community discovery algorithm,visualization layout algorithm and important author recommendation algorithm.Firstly,since the existing co-authorship network community division algorithm is not enough to reflect the characteristics of co-authorship network,a new community division algorithm based on academic community center degree is proposed.Firstly,the algorithm calculates each author's academic community center degree by considering the coherence of the author's relationship strength,then it is applied to the second stage of the Louvain algorithm as the basis for seed nodes selection.The experiment chooses actual co-authorship network data,then visualizes the network structure by using appropriate layout algorithm.Experiment result shows that the improved algorithm can better find the academic communities,and the results are more legitimate.Secondly,since the existing co-authorship network visualization layout algorithm poorly display the community structure,and it's not enough to reflect the characteristics of co-authorship network,a new visualization layout algorithm combine the characteristics of co-authorship network is proposed.Based on the FR layout algorithm,the algorithm optimizes the gravitational and repulsive models by adding the relationship strength and the academic community center degree.In addition,the community gravity force is added to the academic community to achieve the internal clustering.Experiments including comparison with the previous co-authorship network visualization method show that,the improved algorithm has more obvious community layout effect,clearer community structure display,and higher algorithm efficiency.Thirdly,since the recommend author of existing authority evaluation method in co-authorship network usually has a low vitality,an improved important author recommendation algorithm is proposed.On the basis of the Page Rank algorithm,the value ofthe author's signature order is deeply excavated,and the difference of paper published time is used as time factor to reflect the academic activity of authors.The experiment shows that the improved algorithm effectively solves the problems in co-authorship network authority evaluation method,and the important authors' recommendation results are scientific and reasonable.