[目的/意义]在美国对华科技和人才封锁政策背景下,虚拟科研团队发现能够从情报研究的角度,为潜在合作对象识别、具有实体化潜力的科研团队发掘、团队组建及其结构优化提供参考和关键数据支撑,具有重要的理论和实践应用价值。[方法/过程]首先,研究选取Scopus数据库中2012-2021年重症医学领域获NIH资助的研究论文、会议论文和综述,引入并改进社团识别研究中基于模块度的Louvain算法,依据特定交叉科学领域科研人员科研产出合著网络中作者合作的紧密程度,初步识别虚拟科研团队;然后,结合向量空间模型、模拟退火算法和并行计算,进一步优化作者合作关系表征和精炼算法迭代过程,通过优化矩阵计算和矩阵输入提升虚拟科研团队识别的效果和效率;最后,对比GN算法和谱聚类算法的识别结果,验证改进的Louvain算法识别虚拟科研团队的有效性。[结果/结论]在大规模作者-作者关系矩阵运算中,将改进的Louvain算法用于识别特定领域的虚拟科研团队较为理想,具体表现在以下3个方面:①充分考虑情报学研究中作者对论文的贡献,提高科研团队识别效果;②通过优化矩阵输入,精炼局部模块度计算;增加并行计算,有效减少运算时间,提升识别效率;③虚拟科研团队识别结果为潜在合作对象的发现提供参考。
[Purpose/Significance] Under the background of the US technology against China and talent blockade policy, the virtual scientific research team finds that from the perspective of intelligence research, it provides references and key data support for identifying potential cooperation objects, discovering scientific research teams with substantial potential, and forming a team and optimizing the structure, which has important theoretical and practical application value. [Method/Process] This study selected NIH-funded research papers, conference papers and reviews in the field of critical care medicine from the Scopus database from 2012 to 2021, introduced and improved the modularity-based Louvain algorithm in community identification research, and based on the closeness characteristics of author cooperation in the scientific research output co-authorship network of scientific researchers in a specific cross-scientific field, this paper initially identified of virtual scientific research teams; Then, combined with vector space model, simulated annealing algorithm and parallel computing, respectively, it further optimized the iterative process of author partnership characterization and refinement algorithm, through optimization matrix calculation and matrix input, it improved the effect and efficiency of virtual scientific research team identification; Finally, the identification results of GN algorithm and spectral clustering algorithm were compared to verify the effectiveness of the improved Louvain algorithm to identify virtual scientific research teams. [Result/Conclusion] In the calculation of large-scale author-author relationship matrix, it is ideal to use the improved Louvain algorithm to identify virtual scientific research teams in specific fields, it is manifested in the following three aspects. First, the contribution of authors to papers in information science research is fully considered, to improve the identification effect of the scientific research team. Second, by optimizing the matrix input, refining the local modularity calculation and increasing the parallel calculation, the operation time is effectively reduced, and the recognition efficiency is improved. Third, the identification results of the virtual scientific research team provide a reference for the discovery of potential partners.
[1] FINANCESONLINE. 12 Virtual team trends for 2021/2022: top forecasts to watch out for [EB/OL].[2022-03-02]. https://financesonline.com/virtual-team-trends/.
[2] UPWORK. Third annual "Future Workforce Report" sheds light on how younger generations are reshaping the future of work [EB/OL].[2022-03-10]. https://www.upwork.com/press/releases/third-annual-future-workforce-report.
[3] 渠慎宁, 杨丹辉. 逆全球化下中美经济脱钩风险的领域与应对策略[J]. 财经问题研究. 2021(7):102-109.
[4] LI J Y, WAN X X, WANG X. Effects of social capital and knowledge integration on innovation performance: an example of virtual teams[J]. Revista de cercetare si interventie sociala, 2020, 69: 227-240.
[5] STOKOLS D, HALL K L, TAYLOR B K, et al. The science of team science: overview of the field and introduction to the supplement[J]. American journal of preventive medicine, 2008, 35(2S): 77-89.
[6] NATIONAL RESEARCH COUNCIL. Enhancing the effectiveness of team science [EB/OL].[2022-03-20]. http://www.nap.edu/catalog/19007/enhancing-the-effectiveness-of-team-science.
[7] JESUS R, BELEN G, JOSE M. Scientists' performance and consolidation of research teams in biology and biomedicine at the spanish council for scientific research[J]. Scientometrics, 2006, 69(2): 183-212.
[8] BORDONS M, ZULUETA M A, CABRERO A, et al. Identifying research teams with bibliometric tools[EB/OL].[2022-01-12].https://pascal-francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=3145915.
[9] 邱均平, 伍超. 基于社会网络分析的国内计量学作者合作关系研究 [J]. 图书情报知识, 2011(6): 12-17.
[10] CALERO C, BUTER R, VALDES C, et al. How to identify research groups using publication analysis: an example in the field of nanotechnology[J]. Scientometrics, 2006, 66(2): 365-376.
[11] 吕璐成, 赵亚娟, 王学昭, 等. 基于关联规则挖掘的研发团队识别方法 [J]. 科技管理研究, 2016, 36(17): 148-152.
[12] 余厚强, 白宽, 邹本涛, 等. 人工智能领域科研团队识别与领军团队提取 [J]. 图书情报工作, 2020, 64(20): 4-13.
[13] VINCENT B, JEANLOUP G, RENAUD L, et al. Fast unfolding of communities in large networks[J]. Journal of statistical mechanics: theory and experiment, 2008, 2008(10): 1-12.
[14] NEWMAN M E J, GIRVAN M. Finding and evaluating community structure in networks[J]. Physical review e statistical nonlinear & soft matter physics, 2004, 69(2): 026113.
[15] SATTARI M, ZAMANIFAR K. A cascade information diffusion based label propagation algorithm for community detection in dynamic social networks[J]. Journal of computational science, 2018, 25: 122-133.
[16] WANG M, ZOU Y, CAO Y, et al. Searching software knowledge graph with question[EB/OL].[2022-01-12].https://linkspringer.53yu.com/chapter/10.1007/978-3-030-22888-0_9.
[17] 杜伟静, 李翀, 王宇宸, 等. Web of Science科研社区挖掘算法研究 [J]. 小型微型计算机系统, 2020, 41(12): 2465-2469.
[18] NEWMAN M E J. Modularity and community structure in networks[J]. Proceedings of the National Academy of Sciences, 2006, 103(23): 8577-8582.
[19] 方勇, 杨京宁, 颜佳佳, 等. 自然指数和基本科学指标在基础研究领域影响力的差异化分析 [J]. 科技管理研究, 2017, 37(7): 56-60.
[20] 杨倩倩, 刘宪, 马德章. 自然指数对高校科研能力评估的意义与思考——以南方科技大学为例 [J]. 科研管理, 2020, 41(7): 258-261.
[21] 杨颖, 许丹, 陈斯斯, 等. 基于自然指数刊文数据对全球医学研究领域热点的探析 [J]. 情报学报, 2019, 38(11).1129-1137.
[22] NATURE. Nature Index 2014 [EB/OL].[2021-12-17]. https://www.natureindex.com/faq.
[23] NATURE. Nature Index tables [J]. Nature, 2015, 522(7556S): 34-44.
[24] 沈耕宇,黄水清,王东波.以作者合作共现为源数据的科研团队发掘方法研究[J].现代图书情报技术,2013(1):57-62.
[25] 孙登第. 基于随机点积图理论的模式识别方法研究[D]合肥:安徽大学, 2012.
[26] 李宝强, 李翠萍, 张琳, 等. 基于分段加权的点积相似度方法研究 [J]. 计算机与应用化学, 2014, 31(1): 24-28.
[27] 何嘉林. 复杂网络中的社团结构探测和应用研究 [D]. 成都:电子科技大学, 2017.
[28] 王丰雪, 陈家琪. 一种结合模拟退火和贪心策略的社团识别算法 [J]. 电子科技, 2016, 29(2): 8-11.
[29] METROPOLIS N, ROSENBLUTH A, ROSENBLUTH M, et al. Equation of state by fast computing machines[J]. Journal of chemical physics, 1953, 21(6):1087-1092.
[30] KIRPATRICK S, GELATT C D, VECCHI M P. Optimization by simulated annealing[J]. Science, 1983, 220(4598):671-680.
[31] 中华医学会重症医学分会. 中国重症医学科研平台章程[EB/OL].[2022-03-02]. http://www.csccm.org/cn/page.asp?pageid=119.html.
[32] LIAO X, WANG B, KANG Y. Novel coronavirus infection during the 2019-2020 epidemic: preparing intensive care units-the experience in Sichuan Province, China[J]. Intensive care medicine, 2020, 46(2): 357-360.
[33] 李纲, 柳明飞, 吴青, 等. 基于蝴蝶结模型的科研团队角色识别及其特征研究 [J]. 图书情报工作, 2017, 61(5): 87-94.
