基于关键词特征和多元异构信息的论文颠覆性指数预测模型研究  

Study on Predictive Model of Paper Disruptive Index Based on Keyword Features and Multi-source Heterogeneous Information

作　　者：李欣哲 鲁晓 Li Xinzhe;Lu Xiao(Institutes of Science and Development,Chinese Academy of Sciences,Beijing 100190;School of Public Policy and Management,University of Chinese Academy of Sciences,Beijing 100149)

机构地区：[1]中国科学院科技战略咨询研究院,北京100190 [2]中国科学院大学公共政策与管理学院,北京100149

出　　处：《情报探索》2025年第2期1-10,共10页Information Research

基　　金：国家社会科学基金重大项目“世界科技强国制度环境的比较研究”(项目编号:23&ZD149);国家自然科学基金委软课题“新形势下自然科学基金的新定位及应用基础研究的资助体系探究”(项目编号:L2424113)研究成果之一。

摘　　要：[目的/意义]旨在克服传统评价论文方法的滞后性和主观性。[方法/过程]从论文标题和摘要中提取关键词特征,并融合学科领域、参考文献数量、作者机构类型等多源信息,基于随机森林算法提取的特征,构建遗传算法优化的深度神经网络预测模型。[结果/结论]该模型为学术评价提供了新的视角和方法,能够对新发表论文作出及时的颠覆性评估,为学者研判领域发展趋势、编委会审稿决策、基金资助评审等提供支持。未来工作将进一步挖掘相关特征,提高模型的预测性能。s of the papers,and integrates multi-source information such as subject areas,number of references,and types of authors’organizations.Based on the features extracted by the random forest algorithm,this paper constructs a deep neural network prediction model optimized by genetic algorithm.[Result/conclusion]The model provides a new perspective and methodology for academic evaluation,and is able to make timely disruptive assessment of newly published papers,which can support scholars’research on the development trend of the field,editorial boards’decision-making on reviewing manuscripts,and evaluation of fund grants.The next study will further explore the relevant features to improve the prediction performance of the model.

关 键 词：颠覆性指数 深度神经网络 知识创新 机器学习 成果预测 

分 类 号：G353.1[文化科学—情报学] G316