颠覆性技术的后向科学引文知识特征识别研究——以基因工程领域为例  

Research on the Knowledge Feature Identification of Disruptive Technologies from Its Backward Scientific Citations: Taking the Field of Genetic Engineering as an Example

作　　者：马亚雪 王嘉杰 巴志超 孙建军[1,2] Ma Yaxue;Wang Jiajie;Ba Zhichao;Sun Jianjun(Laboratory for Data Intelligence and Interdisciplinary Innovation,Nanjing University,Nanjing 210023;School of Information Management,Nanjing University,Nanjing 210023;Research Institute for Data Management Innovation,Nanjing University,Suzhou 215011)

机构地区：[1]南京大学数据智能与交叉创新实验室,南京210023 [2]南京大学信息管理学院,南京210023 [3]南京大学数据管理创新研究中心,苏州215011

出　　处：《图书情报工作》2024年第1期116-126,共11页Library and Information Service

基　　金：国家自然科学基金青年项目“基于特征挖掘的科学问题域创新状态建模与突破机理研究”(项目编号:72204109);江苏省社会科学基金青年项目“战略性新兴技术领域知识涌现模式与创新态势探测研究”(项目编号:22TQC003)研究成果之一。

摘　　要：[目的/意义]颠覆性技术对国家的政治、经济与社会安全起到重要的支撑作用。准确识别颠覆性技术特征,对于预见技术创新发展方向、优化国家创新战略布局具有重要意义。[方法/过程]聚焦专利引证科学论文(即后向科学引文)的知识特征对颠覆性技术的预见能力,首先,从被引科学论文的知识内容特征和施被引文献的知识关联特征两个方面,量化专利后向科学引文的知识特征;然后,采用逐步回归法初步识别与专利创新程度具有显著关联的候选特征;最后,引入机器学习算法与SHapley Additive exPlanations(SHAP)模型深度解析后向科学引文知识特征与专利创新程度的关联关系,进而识别具有颠覆性技术预测能力的专利后向科学引文知识特征。[结果/结论]基因工程领域的实证分析结果显示,除被引论文的知识组合新颖性外,其余后向科学引文知识特征均与专利创新程度存在显著的非线性关联关系。在一定阈值范围内,颠覆性技术的后向科学引文可能表现出较低的知识影响力和知识元多样性、较强的知识跨学科性,并且与施引专利之间存在较长的发表时间间隔和较大的知识内容相关性。[Purpose/Significance]Disruptive technology plays an important supporting role in the political,economic,and social security of a country.Accurately identifying the features of disruptive technologies is of great significance for predicting the direction of technological innovation and optimizing the national innovation strategic layout.[Method/Process]This paper focuses on the ability of the knowledge features of patents’scientific references(i.e.,backward scientific citations)to foresee disruptive technologies.Firstly,the knowledge features of backward scientific citations are quantified from two aspects:the knowledge content of the cited scientific papers and the knowledge association between citing and cited literature.Then,the stepwise regression method is used to preliminarily identify the candidate features significantly related to the degree of patent innovation.Finally,machine learning methods and the Shapley additional explanations(SHAP)model are introduced to deeply analyze the correlation between the knowledge features of backward scientific citations and the degree of patent innovation,and thus identify the critical knowledge features of backward scientific citations that can predict disruptive technologies.[Result/Conclusion]The empirical analysis results in the field of genetic engineering indicate that,except for the novelty of cited papers’knowledge combination,there is a significant nonlinear correlation between other knowledge features of scientific references and the degree of patent innovation.Within a certain threshold range,the knowledge features of disruptive technologies’backward scientific citations may exhibit low knowledge impact and knowledge element diversity,high knowledge interdisciplinary,as well as long publication time interval and great knowledge correlation with citing patents.

关 键 词：颠覆性技术 后向科学引文 特征识别 基因工程 

分 类 号：G250[文化科学—图书馆学]