Community Structure Of Weighted Innovation Network And Its Effect On Innovation Performance

Innovation agglomeration is conducive to promote the emergence of innovation.Highly concentrated knowledge spillover and frequent interactive learning among innovators make it possible for a large number of innovative achievements to emerge.With the development of economic globalization,information networking and knowledge industrialization,the trend of innovation agglomeration has changed from geographical agglomeration to network agglomeration.At present,the research on innovation agglomeration mostly studies the impacts of agglomeration on innovation from the perspective of space.It is urgent to enrich the research on innovation agglomeration from the perspective of innovation network.The phenomenon of innovation agglomeration in innovation network is manifested as community structure.Studying innovation agglomeration from the perspective of community structure can better describe the current form of network agglomeration.And it is also the future direction of research.Firstly,this paper summarized the current research progress on innovation networks and community detection.Research issues were put forward based on the shortcomings of the existing researches.Secondly,combining with partnerships among innovators,a weighted innovation network model was constructed based on the cooperative frequency.Thirdly,community structure in weighted innovation network was defined.And a fast modularity optimization algorithm was used to detect the community structure.On the basis,the evolution of community structure was studied.Then,the influence mechanism of community structure on innovation performance was analyzed from perspectives of innovation resource agglomeration,knowledge spillover effect and innovation risk reduction.Finally,five research hypotheses were put forward based on the mechanism analysis and relevant scholars' researches.And taking the community data as samples,an empirical study was made to verify the research hypotheses by using negative binomial regression model.The main conclusions were as follows:(1)The weighted innovation networks of China's new energy automobile industry and high-end equipment manufacturing industry have obvious scale-free and small-world characteristics.(2)The average existence time of large communities is longer than that of small communities.And their nodes change more dynamically.For small communities,their existence time will be longer if the change rate of nodes is low.Otherwise,their existence time will be shorter.For large communities,their existence time will be longer if the change rate of nodes is high.(3)Strong ties,number of communities,average size of communities,average existence time of communities and average change rate of communities all have positive impacts on the improvement of innovation performance.The research shows that the weighted innovation network model can better describe the real innovation network.And the community structure can truly reflect the phenomenon of innovation agglomeration in innovation network.From the perspective of community structure,we can deepen our understandings of network topology,innovation agglomeration and its impact on innovation performance.Furthermore,we can also reveal the dynamic change law of community structure and the mechanism of promoting innovation.Based on these findings,decision-making references for promoting innovation agglomeration and upgrading the level of industrial innovation are propounded.