摘要
[目的/意义]精准识别创新前沿有利于国家、政府、企业对创新战略进行前瞻性部署,对于抢占技术先机、赢取竞争优势具有积极意义。[方法/过程]首先构建机器学习模型,通过预测近期发表的论文被高度引用的概率识别潜在高被引论文,同时基于技术新颖性、技术独特性、技术重要性3个维度,构建一套评价技术创新水平高低的指标体系来筛选高价值专利;然后采用LDA主题模型分别对潜在高被引论文和高价值专利进行聚类分析,识别科学创新前沿、技术创新前沿、科技创新前沿;最后根据创建的科学价值、技术价值指标,结合主题强度构建创新前沿地图,量化解读创新前沿之间的发展水平及价值差异。[结果/结论]以智能驾驶汽车领域进行实证研究表明,该方法可以有效识别创新前沿,并能够展现创新前沿之间的科学价值、技术价值、主题强度差异,能够为国家、企业的技术布局、策略制定提供参考。
[Purpose/Significance]Accurately identifying the innovation fronts is conducive to the forward-looking deployment of innovation strategies by the state,the government and enterprises,and is of positive significance for seizing technological opportunities and winning competitive advantages.[Method/Process]Firstly,build a machine learning model to predict the probability that recently published papers are highly cited and identify potential highly cited papers.At the same time,build a set of index system to evaluate the level of technology innovation and screen high value patents based on the three dimensions of technological novelty,technological uniqueness and technological importance.Then,LDA theme model was used to cluster the potentially highly cited papers and high-value patents respectively,so as to identify scientific innovation fronts,technological innovation fronts and scientific-technological innovation fronts.Finally,according to the created scientific value and technical value index,combined with the theme intensity,build a map of the innovation fronts,and quantitatively interpret the development level and value differences between the innovation fronts.[Result/Conclusion]The empirical research based on intelligent driving vehicle data shows that this method can effectively identify the innovation fronts,show the scientific value,technical value and theme intensity differences between the innovation fronts,and provide references for the technical layout and strategy formulation of countries and enterprises.
作者
张彪
吴红
高道斌
林艳秋
Zhang Biao;Wu Hong;Gao Daobin;Lin Yanqiu(Institute of Information Management,Shandong University of Technology,Zibo 255049)
出处
《图书情报工作》
CSSCI
北大核心
2022年第18期72-83,共12页
Library and Information Service
基金
国家社会科学基金项目"高校图书馆深度嵌入专利运营研究"(项目编号:16BTQ029)研究成果之一。
关键词
潜在高被引论文
高价值专利
创新前沿
机器学习
文本主题聚类
potentially highly cited paper
high value patent
innovation front
machine learning
text topic clustering
