摘要
[研究目的]新质生产力的发展极大程度上依赖于颠覆性技术的突破。准确识别颠覆性技术有助于推动生产能力现代化,增强国家实力和社会发展水平。[研究方法]以专利家族作为技术分析单元,融合专利数据和论文数据,从多样性、均衡性、差异性和Rao-Stirling综合维度挖掘颠覆性技术所引证知识的学科交叉特征,并据此采用逻辑回归算法识别与技术颠覆性程度具有显著关联的候选特征;构建八类机器学习模型并优选颠覆性技术预测效果最佳的模型,通过SHAP模型揭示学科交叉特征在颠覆性技术预测中的相对贡献和特征关联机制。[研究结果/结论]人工智能领域研究结果表明,所引证专利和论文的多样性、均衡性和差异性特征均对颠覆性技术的产生具有显著影响,相较于其他八类机器学习模型,XGBoost模型在综合性能上取得了最佳表现,其中引证论文的差异性、专利的多样性和差异性等交叉驱动特征在颠覆性技术预测中贡献度最高。
[Research purpose]The development of new productive forces largely depends on breakthroughs in disruptive technologies.Accurately identifying disruptive technologies contributes to promote the modernization of production capacity,and enhance national strength and social development.[Research method]Taking patent families as the unit of technical analysis,this study integrates patent and publication data to explore the interdisciplinary characteristics of the knowledge cited by disruptive technologies from dimensions such as diversity,balance,disparity,and the Rao-Stirling diversity index.A logistic regression algorithm is employed to identify candidate features that are strongly associated with the degree of technological disruption.Eight machine learning models are constructed to identify the best-performing model for predicting disruptive technologies.The SHAP model is used to reveal the contributions of interdisciplinary features and their underlying mechanisms in predicting disruptive technologies.[Research result/conclusion]The results in the field of artificial intelligence show that the diversity,balance,and disparity of cited patents and publications all significantly impact the generation of disruptive technologies.Among the eight machine learning models,the XGBoost model achieved the highest predictive accuracy.Among the interdisciplinary-driven features,the disparity of cited publications and the diversity and disparity of patents made the highest contributions to the prediction of disruptive technologies.
作者
王萌萌
吴艾晗
邓琨升
郭晓彤
Wang Mengmeng;Wu Aihan;Deng Kunsheng;Guo Xiaotong(College of Management,Xi'an University of Architecture and Technology,Xi'an 710055;College of Information and Control Engineering,Xi'an University of Architecture and Technology,Xi'an 710055)
出处
《情报杂志》
北大核心
2025年第3期72-80,138,共10页
Journal of Intelligence
基金
国家自然科学基金青年基金项目“学科交叉背景下知识融合的触发及驱动机制研究——以跨学科团队为例”(编号:72104192)研究成果。
关键词
颠覆性技术
学科交叉
专利家族
专利数据
科学论文
人工智能
机器学习
SHAP模型
disruptive technology
interdisciplinary
patent family
patent data
scientific papers
artificial intelligence
machine learning
SHAP model
