Identification Model Construction Of Interdisciplinary Breakthrough Innovation Based On Knowledge Flow Theory

With the advent of"the era of great science",the production of knowledge has long entered the era of multidisciplinary cross-integration,so as to the intersection of disciplines often becomes a new scientific growth point and scientific frontier.The results of a number of case studies in recent years also show that major breakthrough innovation research represented by Nobel Prize-winning achievements has shown an increasing trend in integrating the amount of knowledge from different types of disciplines,and the integration of disciplinary knowledge has increasingly become the main way for scientific research to achieve breakthrough innovation.Therefore,timely and accurate identification of interdisciplinary breakthrough innovation is of great significance to promote the prospective selection and cultivation of interdisciplinary breakthrough innovation research and enhance the ability of source innovation.However,the current research on interdisciplinary breakthrough innovation mainly focuses on the two aspects of influencing factor research and cooperative behavior and its interaction mechanism,and few studies take interdisciplinary breakthrough innovation as the research object to analyze its typical characteristics and construct identification models accordingly.Moreover,although many scholars have evaluated breakthrough innovation from different angles in the past,these traditional studies are only aimed at the performance of breakthrough innovation,and most of them only consider the external characteristics of the text,so it is difficult to make a deep analysis of the production mechanism from the content characteristics of the cited scientific knowledge.In addition,interdisciplinary breakthrough innovation and breakthrough innovation have different knowledge production mechanisms,thus the previous identification indicators and methods are not fully applicable to the identification research of interdisciplinary breakthrough innovation.In view of the above problems,based on the theory of knowledge flow,we first analyzed the internal mechanism of interdisciplinary breakthrough innovation.Then on this basis,we divided the research process of interdisciplinary breakthrough innovation into three stages:interdisciplinary knowledge integration,interdisciplinary knowledge absorption and interdisciplinary knowledge diffusion,and designed relevant metrics by comprehensively analyzing the characteristics and performance of different dimensions of knowledge integration at each stage.And we set the interdisciplinary breakthrough innovation research that won the Nobel Prize as experimental data,including all interdisciplinary researches that won the Nobel Prize in Physics,Chemistry and Physiology and Medicine from 1901 to 2021.We also set up a control group,that is,articles published in the same year,in the same journal and same research fields as the key papers that won the Nobel Prize.Then,on the basis of considering the influence of discipline factors and time factors on the indicators,through the comparative analysis of the experimental group and the control group,including the overall characteristics analysis and significance level analysis of the indicators under the three dimensions of interdisciplinary knowledge integration,interdisciplinary knowledge absorption and interdisciplinary knowledge diffusion,the typical characteristics of interdisciplinary breakthrough innovation were extracted.And we conducted multicollinearity analysis and correlation analysis to determine the final input indicators.Then,based on these typical characteristic indicators,the Logistic regression analysis method was used to construct an interdisciplinary breakthrough innovation identification model,and the optimal citation time window was also determined by cross-validating the models with different citation time windows by fivefold,so as to reduce the possible bias of premature or too late evaluation of high-quality results.Finally,300 interdisciplinary general papers were selected in the interdisciplinary research fields of Physics,Chemistry and Physiology and Medicine,and used as test data sets for empirical research together with 133 existing Nobel Prize-winning interdisciplinary breakthrough innovation research.Then,the constructed identification model was applied to the datasets in these fields to identify the interdisciplinary breakthrough innovation research that won the Nobel Prize.By statistically calculating the matching accuracy between the model identification category and the real category in the final identification results,the performance of the identification model was effectively evaluated.Finally,by comparing the identification results with those based on the disruptive index DI and its improvement index DI₅,it was found that the recognition accuracy of the interdisciplinary breakthrough innovation identification model constructed in this study is better than the overall recognition effect based on the disruptive index DI and its improvement index DI₅,and the model has strong interdisciplinary generalization ability and stability.It is appropriate to set the optimal citation window to 3 years,which will facilitate the early identification and cultivation of interdisciplinary breakthrough innovations.