负联系对创新网络结构演化的影响

The Effect of Negative Links:How Innovation Network Evolves

结合结构平衡理论与动态平衡思想,对企业创新网络演化过程中负联系的影响和作用机制进行深入分析,界定了负联系影响下创新个体的4种知识搜索行为和2种搜索屏蔽现象,并分析了知识搜索行为演化机制。研究结果表明:①负联系会缩小创新个体间地位差距,使个体地位趋于平等;②负联系会使创新网络始终保持鲜明的派系特征,并阻碍派系融合,使创新个体更依赖于其所在派系;③创新网络凝聚性不仅取决于个体互动中建立的正联系,还受到负联系的复杂影响。一方面,负联系通过抑制网络正联系、增强拓扑性削弱创新网络整体的小世界现象;另一方面,负联系加深了创新个体对派系的依赖程度,使得派系内成员间的联系更紧密,网络局部小世界结构特征更显著。

Combining the structural balance theory and the dynamic balance theory,this paper focuses on the influence mechanism of negative links of the inter-organizational innovation network structure.To overcome the difficulty of the ac-quisition of empirical data,the paper conducts an agent-based modeling and simulation research.An inter-organizational innovation network is composed of a group of firms or organizations whose network behaviors contribute the bottom-up re-sult of network evolution.In order to reduce unnecessary complexity,the paper limits firms'network behavior as firms'knowledge search which is based on the"triadic"network mechanism.Then it identifies four kinds of knowledge search be-havior of innovative agents or firms,two kind of shielding phenomenon and the dynamic characteristics of knowledge search behavior,and finally builds a multi-agent model to simulate inter-organizational innovation network evolution process of innovation network.It is found that at first negative links will fill the status gaps between the individual agents and keep the whole innova-tion network at a high level of equality.Secondly,negative links will be make the whole innovation network keep an ar-resting faction structure and hinder the convergence between factions.Thus,individual agents will be dependent on their factions strongly.At last,different with the previous conclusions,the paper shows that the network cohesion depends not only on the well-known positive links which represent knowledge sharing and knowledge collaboration,but also on nega-tive links which represent hostility,confrontation and knowledge blocking.On the one hand,negative links will suppress the increasing trend of innovation network's positive topology and weaken the small-world phenomenon of the whole net-work.On the other hand,under the influence of negative links,the individual agent increases the reliance on its own frac-tion,which leads to closer contact between individual agents within factions.The theoretical contribution of this paper is reflected in three aspects.First this study uses the multi-agent modeling and simulation methods to study the impact of negative connections on the evolution of innovation network structure,pro-viding new research ideas and research perspectives for the introduction of negative connections or innovation networks un-der the framework of symbolic networks.Secondly,it defines enterprise knowledge search behavior as"knowledge tracea-bility search","knowledge source sharing search","diffusion source sharing search"and"alliance search"by introducing the idea of structural balance and dynamic balance of symbolic networks,and enriches knowledge search research.Third-ly,on the basis of the four types of knowledge search behaviors,this study builds a multi-agent simulation model for the evolution of inter-enterprise innovation network structure under the influence of negative connections,which enriches the application of multi-agent modeling and simulation methods in innovation networks,social networks and complex net-works.The limitations of this paper are reflected in the following two aspects.First of all,there are many typical network behaviors of innovative individuals,and each basic type governs multiple seed classes.In this study,the network behavior of innovation individuals is limited to the"knowledge search behavior"based on the network transmission mechanism,which will lead to the universality of the multi-agent simulation model of innovation symbol network established on this basis to a certain extent.Secondly,for the multi-agent modeling method itself,although it can overcome the difficulties in obtaining empirical research data to a certain extent,the application of this method in the field of innovation networks has just started,and there is still much room for progress in the corresponding aspects of the model and reality,mainly reflec-ted in that the individual behavior in the simulation model can not be well connected with the actual network behavior of enterprises.The individual behavior described by computer language is too simple and abstract,while the real network be-havior of enterprises is more complex and specific.This makes the relevant conclusions difficult to understand.In addi-tion,network structure is the embodiment of network functions and performance,and the negative relationship that has an important impact on the evolution of innovation network structure is not involved in this study.Therefore,in the future research,it is necessary to carry out a deeper expansion in the innovation of individual network behavior and the corre-sponding relationship between network structure and network performance.