基于结构参数的科研合作网络知识扩散建模研究

Modeling Knowledge Diffusion in Scientific Collaboration Network Based on Structural Parameters

科研合作网络上的知识传播扩散现象往往同时受到多种网络结构因素的共同影响。为了揭示科研合作网络结构与知识扩散过程的关系，本文从全局、局部以及个体三个层次，选取网络平均路径、派系数、聚集系数等13项科研合作网络结构测度参数，提取科研合作网络结构因子，将知识负载、知识多样性以及知识冗余度作为衡量知识扩散效果的评价指标，利用多元线性回归模型建立科研合作网络结构与知识扩散效果之间的定量关系模型，即科研合作网络知识扩散结构参数模型，并对模型进行验证，以探索科研合作网络结构对知识传播扩散的影响机制。研究表明：“知识借助科研合作关系进行传播扩散，且科研合作网络结构影响知识的扩散进程”的前提假设是正确的；本文所构建的科研合作网络知识扩散结构参数理论模型是合理的；从科研合作网络结构因子粒度来看，科研合作网络距离因子以及聚集因子对知识负载、知识冗余度和知识多样性都起着正向促进的作用，模块化因子对知识负载和知识冗余度起着正向促进的作用，但对知识多样性几乎不产生显著影响；从科研合作网络参数的粒度来看，不同结构参数对知识扩散效果的影响既存在联系又有一定差异；知识的传播与扩散同时受到科研合作网络结构各参数（或因子）的共同影响。

Knowledge diffusion in scientific collaboration network tends to have combined impact from various network structure factors simultaneously. Thirteen kinds of parameter that represent the structure of, scientific collaboration network from the global, local and individual levels are elected to further extract the structural factors of scientific collaboration network. With knowledge load, knowledge diversity and knowledge redundancy functioning as the evaluation indicators to measure the effect of knowledge diffusion, the structural parameter quantitative relationship model of multivariate linear regression of knowledge diffusion in scientific collaboration network is formed and verified, so as to explore the mechanism by which the scientific collaboration network structure affect knowledge diffusion. The research has shown that： Firstly, The hypothesis that knowledge can further be diffused with the aid of scientific collaboration and the scientific collaboration network structure will influence the process of knowledge diffusion are justified, and the structural parameter quantitative relationship model of multivariate linear regression of knowledge diffusion in scientific collaboration network is rational and reasonable. Secondly; from the perspective of the granularity of structural factors in scientific collaboration network, the distance factor and clustered factor of scientific collaboration network has positive effect on knowledge load, knowledge redundancy and knowledge diversity, while modular factor has positive effect on knowledge load and knowledge redundancy, but no conspicuous effect on knowledge diversity. Thirdly, from the perspective of the granularity of parameter in scientific collaboration network, the influence of different kinds of structural parameter on the effect of knowledge diffusion has both relevance and differences. Fourthly, different kinds of structural parameter or factors have joint influence on the knowledge transmission and diffusion simultaneously.