科研合作网络中团队成员对创新团队演化的影响:以机器学习领域为例  

On the Influence of Group Members to Research Group Evolution in Scientific Collaboration Networks:Case of the Machine Learning Field

作　　者：赵伟[1] 夏昊翔[1] ZHAO Wei;XIA Haoxiang(School of Economics and Management,Dalian University of Technology,Dalian 116024,China)

机构地区：[1]大连理工大学经济管理学院,辽宁大连116024

出　　处：《大连理工大学学报（社会科学版）》2022年第1期60-68,共9页Journal of Dalian University of Technology(Social Sciences)

基　　金：国家自然科学基金面上项目“协同创新网络中的集体智能动态机理研究”(71871042);国家自然科学基金重点项目“大数据环境下知识融合与服务的方法及其在电子政务中的应用研究”(71533001);教育部人文社会科学规划基金项目“综合数据解析和动力学建模的科研协作系统演化模式与机理研究”(18YJA630118)。

摘　　要：以上海交通大学发布的学术知识图谱(AceKG)作为数据源,运用优化过关系衰退的派系过滤法(CPM)发现时分科研合作网络中的创新团队演化路径。提出团队在演化过程中的7种事件及团队中成员的3种分类,经过分析发现:短期参与者是影响团队规模的重要因素,团队中的持续参与者与团队生存期相关。文章最后提出了一种利用移动平均法预测团队消亡的方法。根据以上发现,科研工作者为了更好地完成研究工作,可以根据工作的性质,有目的地选择成员,达到控制团队规模和团队的生存期趋势的目的。The Acemap knowledge graph published by Shanghai Jiaotong University was used as the dataset to optimize the group evolution path by clique percolation method in discovering temporal scientific collaboration network.Seven types of events and three categories of group members in the group evolution were proposed.The results show that short-term members were an important factor to affect group size while persistent participants were related to group lifespan.Finally,a Moving-Average-based method was proposed to predict the group dissolution.According to the above findings,researchers can select members according to the nature of work,control the group size and promote the group lifespan in order to better fulfill research goals.

关 键 词：派系过滤法 关系衰退 演化路径 移动平均 

分 类 号：G311[文化科学]