现代情报2025，Vol.45Issue(10)：100-114,15.DOI:10.3969/j.issn.1008-0821.2025.10.009
基于机器学习的科研合作持续性预测:维度解析与模型构建
Prediction of the Scientific Collaboration Sustainability Through Machine Learning:Dimensional Analysis and Model Construction
摘要
Abstract
[Purpose/Significance]To explore the characteristics and rules of the scientific collaboration sustainability is helpful to improve collaboration efficiency and provide decision support for researchers.[Method/Process]This study firstly expounded the rationality of three-dimensional measurement of scientific collaboration sustainability,then designed the indicators of collaboration persistence,collaboration stability and collaboration adhesion respectively,and finally con-structed prediction models of scientific collaboration sustainability based on machine learning algorithm to achieve the accu-rate prediction.[Result/Conclusion]Firstly,compared with the single dimension,the three-dimensional measurement could analyze the characteristics of scientific collaboration sustainability more comprehensively.Secondly,compared with BPNN,SVR,XGBoost,DT and RF,the Self-XGBoost-LSTM prediction method proposed in this study has a higher accu-racy.The relevant results verify the feasibility and applicability of Self-XGBoost-LSTM in the field of scientific collabora-tion sustainability prediction.关键词
科研合作持续性/合作持久性/合作稳定性/合作黏合性/合作预测Key words
scientific collaboration sustainability/collaboration persistence/collaboration stability/collaboration adhesion/collaboration prediction分类
信息技术与安全科学引用本文复制引用
刘晓婷,黄颖,叶冬梅,张琳..基于机器学习的科研合作持续性预测:维度解析与模型构建[J].现代情报,2025,45(10):100-114,15.基金项目
国家自然科学基金项目"科研人员职业生涯的性别差异和影响机理研究:合作、流动与学术表现"(项目编号:71974150). （项目编号:71974150）
