Research On Two-boundary Network Model Based On Patent Data Set

With the advent of the knowledge economy,the researches on the application and output of scientific knowledge to technological innovation and industrial production have always received much attention.Effectively revealing the knowledge flow between science,technology and productivity can help the researchers have a clear understanding of the process of knowledge application and output,and is of great significance and practical value for the formulation of relative management policies and the rational allocation of various resources.As a carrier of the technological innovation,patent is an important link between scientific knowledge and industrial productivity.Utilizing patent citation links to scientific literature and transfer links to enterprise to explore the process of knowledge flow and transformation is a common method in existing researches.However,scholars usually only focus on a single connection,and fail to clearly describe the complete knowledge flow path to explore the application and transformation of knowledge advances.This paper takes the knowledge flow in the patent network as the research and application background,and applies the method of social network analysis to make full use of the scientific literature-patent-enterprise connections,and proposes a two-boundary network analysis model with knowledge application boundary and knowledge output boundary,which effectively describes the knowledge flow network between science-technology-productivity.At the same time,this paper applies the methods of mathematical statistics and regression analysis to explore the efficiency of knowledge utilization and patent impact based on the real patent data set.At first,the definition and construction of the two-boundary network model are given.The feasibility of the model is analyzed theoretically.In the mean time,two indexes of the distances between the patent and the two boundaries are proposed to measure the efficiencies of knowledge application and knowledge output.Next,using the real patent data set,the validity of the model is experimentally verified.Then,applying the two-boundary network model,statistically analyses of patents in different technical classes,regions,institutions and structural patterns are investigate.Also,we use the negative binomial regression model to quantitatively explore the relationship between patent impact and the distances to the boundaries of knowledge application and knowledge output.Furthermore,from the perspective of knowledge spreading,this article proposes some policy recommendations to promote the flow and transformation of knowledge advances from the views of government policy makers,universities and research institutes and business managers.The research results show that patents in the field of basic natural sciences can utilize the scientific knowledge more efficiently,such as Chemistry,Physics,Electronics,etc..Developed countries pay more attention to knowledge application,while developing countries and regions pay more attention to knowledge output.The closer the patent is to the boundaries of knowledge application and knowledge output,the greater influence of it in the path of knowledge flow.The innovations of this paper are:(1)Considering the patent citation relationship and transfer relationship in the patent network at the same time,a two-boundary network model is proposed based on knowledge flow,which effectively describes the knowledge flow between science-technology-productivity.(2)Based on the shortest distances from the patent to the boundaries of knowledge application and knowledge output,two distance metrics are proposed,which represent the efficiencies of knowledge application and knowledge output respectively.Statistical analyses of patents are conducted from different technical classes,regions,institutions and structural patterns levels.Using the negative binomial regression model to quantitatively investigate the relationship between the shortest distances of patent to two boundaries and patent impact.(3)The two-boundary network model proposed in this paper provides a new theoretical analysis framework for studying the connections between heterogeneous nodes.