Identification And Prediction Of The Field's Scientific Breakthrough Based On Entropy Theory

Innovative development is the main point of strategic development among the world's economic entities.Innovation means a change of paradigm,the liberation of labor,and the improvement of production efficiency,so as to realize the comprehensive progress of social economy,culture and technology.From the perspective of the strategic deployment of various countries and regions,innovation breakthrough is the focus of deployment,which contains huge economic value,scientific research value,and national strategic value.Innovation breakthrough,as a type of innovation,has many characteristics such as non-linearity,innovation,and influence.It is difficult to identify and predict since its features are difficult to quantify.If we can identify and predict innovation breakthrough,we can first optimize the allocation of scientific research funds,rationally dispatch scientific researchers,and invest in important scientific researches.We will be the first to achieve scientific breakthroughs in key domains and complete technological transformation,occupying the technological highlands.As an important carrier of scientific innovation,scientific papers are important data sources for innovation breakthrough researches.Identifying innovation breakthrough by interpreting the content of scientific papers helps to understand the development rules and characteristics of innovation breakthrough,and is of great significance for subsequent related researches and management decisions.At present,the existing research methods of innovation breakthrough are mostly carried out from a relatively single dimension such as innovation,novelty,or interdisciplinarity,that is,most of them use one specific feature as a breakthrough to measure,which ignores the basic development characteristic of knowledge,dynamic development.This research takes scientific papers as the main research object,does not take the technical content into consideration,takes the scientific breakthrough theme as the research object,and aims at the within-domain scientific research,trying to identify and predict such topics.The co-occurrence network of the subject words is mainly used as a proxy for knowledge networks,and the research entry point is that the scientific innovation breakthrough will "influence" the overall state of knowledge networks.The change of the network state is measured through "structural entropy".Therefore,a set of breakthrough innovation identification and prediction methods based on entropy values is constructed.It can identify breakthrough innovations from the perspective of the overall development of the knowledge network.This research can complement the theoretical research of the current innovation breakthrough researches,and has certain practical application significance.The main contents of this study include the following three aspects:Firstly,it analyzes several concepts that are close to the connotation of the innovation field from two dimensions: qualitative and quantitative.It sorts out the characteristics and deficiencies of identification methods of innovation breakthrough from the existing studies.And then clearly point out the features that need to be paid attention to when recognizing innovation breakthrough.It provides theoretical support for the recognition methods proposed in this study.Secondly,a method system for identifying and predicting scientific breakthrough innovation based on entropy is proposed.The entropy value is an indicator of the state of the knowledge network.The first is to explain the relationship between the structural entropy and the state of the knowledge network,and then to sort out the existing methods of constructing the structural entropy and conclude their advantages and disadvantages.Considering both the composition("nodes" and "edges")of the knowledge network,and the development feature("non-extensiveness")of knowledge network,the structural entropy index of this study is constructed.Later,the research identify and predict breakthrough innovations from macro level(an important time point of structural entropy change),meso level(community that has a large impact on network structure entropy)and micro level(the node that has a large impact on network structure entropy and the node that has a sudden change in network structure entropy influence).Finally,in the empirical stage,scientific papers in the field of "genetically engineered vaccines" were used as research objects.In the recognition phase,it is mainly carried out from the three dimensions of structural entropy,and at the same time,it compares with the existing emerging subject recognition and mutation subject recognition results.The prediction stage is to first construct a new network for prediction through the link prediction method,and then use the meso and micro levels to screen scientific innovation breakthroughs.Finally,the effectiveness of the method is evaluated by experts.The empirical results prove that the scientific innovation breakthrough identification and prediction method based on entropy value proposed in this paper has certain application value in genetic engineering vaccine field,and can identify and predict breakthrough innovation to a certain extent.This paper includes 20 figures and 26 tables.