Discovering Collaboration Network Structures and Dynamics in Big Data

理解科学家个体之间如何相互作用，以及这种相互作用如何影响研究生产力和知识传播，对于理解科研合作的动态非常重要。与此同时，关于合作模式及其后果的信息对科学政策也有影响。在合作网络的定量研究中，出版物合著和引文链接一直是数据的主要来源。然而，作为网络基础设施支持、数据驱动的科学的标志之一，大型数据存储库变得越来越普遍，它们提供了关于科学合作网络的另一种信息来源。该项目调查围绕这样一个国际数据存储库GenBank出现的研究合作网络，并开发数据产品，以支持数据驱动的科学决策和研究。通过利用这一新颖的数据源，该项目为验证和扩展复杂网络理论提供了前所未有的机会，同时产生丰富的数据输出和产品，以支持科学政策研究和政策制定。这项研究填补了2008年科学政策科学路线图（SoSP）确定的一些理论和方法上的空白，特别关注科学合作网络如何形成和发展。这项研究的结果解决了2008年SoSP路线图中缺乏网络分析、可视化分析和科学映射的模型和工具的问题。为了完成该项目所需的数据收集和处理，将开发新的计算程序来解析、提取、存储、转换、拆分、合并和过滤数据；这些将适用于科学政策和创新研究的其他类似数据源的分析。更广泛的影响。通过提供可用的数据集产品原型，该项目将允许研究人员、政策制定者和学生从纵向、专题、地理、机构和作者维度探索GenBank中的研究网络。这些数据集的多维交互式呈现使数据密集型科学政策研究成为可能，并通过过滤、排序、关联和可视化功能支持科学政策制定。数据集和数据产品将通过开放获取机制提供，因此教育工作者、本科生和研究生有充分的机会将这些资源用于教学和研究。雪城大学新成立的数据科学高级研究证书（CAS DS）项目的学生将能够参与该项目，并获得数据收集和处理、数据质量验证、分析和可视化的编程技能。此外，合作网络分析为感兴趣的博士生提供了独立学习或论文研究的机会。研究网络基础设施支持的数据共享和知识扩散的结果有望提高决策者正确评估联邦资助研究成果的能力。

Understanding how individual scientists interact with one another and how such interaction impacts research productivity and knowledge diffusion is important for understanding the dynamics of scientific research collaboration. At the same time, information about patterns of collaboration and their consequences have implications for science policy. In quantitative research on collaboration networks, publication co-authorships and citation-linkages have been the primary source of data. As large data repositories, one of the signposts for cyberinfrastructure-enabled, data-driven science, become increasingly prevalent, however, they offer an alternative source of information about networks of scientific collaboration. This project investigates research collaboration networks emerging around one such international data repository, GenBank, and develops data products to support data-driven science policymaking and research. By utilizing this novel data source the project provides an unprecedented opportunity to validate and expand the theory of complex networks while generating rich data outputs and products to support science policy research and policymaking. This study fills a number of theoretical and methodological gaps identified by the 2008 roadmap for Science of Science Policy (SoSP), with a specific focus on how scientific collaboration networks form and evolve. The outcomes of this study address the lack of models and tools for network analysis, visual analytics, and science mapping outlined in the 2008 roadmap for SoSP. To accomplish the data collection and processing required for this project new computational programs will be developed to parse, extract, store, transform, split, merge, and filter the data; these will be applicable to the analysis of other similar data sources for science policy and innovation research. Broader impacts. By making available dataset product prototypes the project will allow researchers, policy makers, and students to explore research networks in GenBank from longitudinal, thematic, geographical, institutional, and author dimensions. The multi-dimensional, interactive presentations of such datasets enable data-intensive science policy research and support science policymaking through filtering, sorting, associating, and visualization capabilities. The datasets and data products will be made available through an open access mechanism, so educators and undergraduate and graduate students have ample opportunities to use these resources for teaching and research. Students enrolled in Syracuse University's newly established Certificate for Advanced Study in Data Science (CAS DS) program will be able to participate in the project and gain skills in programming for data collection and processing, data quality verification, analysis, and visualization. In addition, the collaboration network analysis provides interested doctoral students an opportunity to do independent study or dissertation research. Findings from studying cyberinfrastructure-supported data sharing and knowledge diffusion is expected to advance policymakers' ability to properly assess the outcomes of federally funded research.