关键词: 创新网络, 网络社群, 社群经纪人, 网络动态, 社群配置, 稳定性, 创新绩效

Abstract:     In innovation activities, inventors have shown a phenomenon of "hugging together", which leads to the clustering of collaborative innovation networks and forms network communities. The network community of inventors is an aggregated structural group in collaborative networks. Inventors within the same community are frequently and closely linked with each other. Community brokers locate on the periphery of a community and have a position advantage of crossing the boundaries of different communities, which are critical to transfer knowledge from outside the community into a form that can be absorbed by inventors within the community. 
    There is a lack of research on how the dynamic and stable structure configuration of different network loci influences innovation. Solving this problem is helpful to provide theoretical guidance for improving innovation capacity and decision supports for innovation network management. Thus, this study aims to explore how network community structure and its dynamic configuration in inventors′ collaborative innovation network influence their innovation capacity in the field of renewable energy. We stress the roles of the overall dynamic of the community and the dynamic of community brokers. Firstly, inventors′ collaborative networks are constructed based on co-inventing relationship in renewable energy patents. Then, inventors′ network communities are identified by adopting GN algorithm. Community dynamic configuration is defined through tracking communities in adjacent phases based on the dynamic difference of network loci between members within communities and community brokers. Four types of community dynamic configuration are defined, and there are independent community, rigid community, volatile community and broking community, respectively. We hypothesize that among the four types of communities, broking communities have the strongest innovation capacity, while volatile communities have the weakest innovation capacity. OLS regression analyses are used to empirically test our hypotheses and the robustness tests are carried out by using nonparametric bootstrap method. 
     Through this study, we aim to break through the limitations of studies on the functional mechanism of innovation networks from the static network perspective and integrate the advantages of static and dynamic networks. We contribute to the study of innovation network structure and network dynamics, enriches studies on the functional mechanism of innovation network, and expect to provide guidance and suggestions for improving inventors′ innovation capacities.
    The results of GN algorithm show that there are obvious community structures in inventors′ collaborative innovation network in the renewable energy field, and network communities and communities′ brokers evolve over time and they present different dynamic configurations. Moreover, the results of OLS regressions and robustness tests show that different dynamic configuration of community has a significantly different impact on community innovation capacity. Overall, the dynamic and static harmonious community configuration is better than the dual dynamic or static community configuration. Specifically, the innovation capacity of the volatile community is weak and that of broking community is strong and that of independent community and rigid community are somewhere in between. This study emphasizes the effect of heterogeneity of community dynamic configuration on innovation of community inventors, and the policy should be conducive to the compromise of dynamic and stable structure of innovation networks, so as to promote innovation of inventors.

Key words:  innovation network, network community, community broker, network dynamic, community configuration, stability, innovation performance