Evolution of hot topics in team science / 团队科学中热点主题的演变

By Ying Huang, Ruinan Li, Yashan Li and Lin Zhang

A Chinese version of this post is available

What are the research hotspots in the Science of Team Science (SciTS) field? How have they evolved in the last decade?

We used conference programs from the annual International Science of Team Science (INSciTS) conferences held between 2010-2019 and the CorTexT Platform (https://www.cortext.net/) to select the top terms used with high frequency in the 852 titles and abstracts.

High-frequency terms and their evolution

The top 25 terms and their evolution are shown in Figure 1: evaluation / assessment (89 mentions), team science training (63), leadership (31), communication (30), curriculum (14),community (29), funding (13), mixed methods (13), model (13), Toolbox Project / Toolbox Dialogue Initiative (13), Clinical and Translational Science Award (CTSA) (11), graduate students (11), networks (26), education (20), tools (20), productivity (16), bibliometrics (19), framework (14), gender (11), readiness (8), distributed collaboration (7), multi-team systems (7), research network tools (8), team performance (7), multiple levels (5).

The two most frequently used terms are “evaluation / assessment” and “team science training.” “Evaluation / assessment” was consistently used with high frequency across the 10 years. The use of “team science training” increased after 2014. The figure also shows peaks and troughs in the use of other terms.

Research topics and their evolution

We then applied the community detection algorithm (Louvain Algorithm) to group the 58 main terms into six ‘communities,’ which are interpreted as five topics (shown in Figure 2):

1. measurement and evaluation of team science,
2. institutional support and professional development for teams,
3. characteristics and dynamics of teams,
4. management and organization for teams, and
5. structure and context for teams.

The formation of two separate clusters for “characteristics and dynamics of teams” may have resulted from the small amount of data. We used the same terms as Falk-Krzesinski et al. (2011) who conducted a concept mapping exercise. Our analysis did not isolate two topics identified by Falk-Krzesinski and colleagues (“definitions and models of team science” and “disciplinary dynamics and team science”), instead these were highly correlated with the other topics.

We further examined the evolution of these research topics as follows.

First, we used the community detection algorithm (Louvain Algorithm) to identify the key research topics in each two-year period across the decade. The topics were labeled using the two nodes with the most frequent co-occurrence in each cluster of nodes, rather than the labels used in figure 2. The topics identified were:

• 2010-2011: (i) community and Team Science Toolkit, (ii) social network analysis and social network, (iii) impact and effectiveness
• 2012-2013: (i) team formation and model, (ii) promotion and tenure, (iii) impact and evaluation/assessment, (iv) team science training and learning, (v) undergraduate students and education
• 2014-2015: (i) outcomes/outputs and measurement, (ii) evaluation/assessment and translational research, (iii) impact and productivity, (iv) team science skills and learning, (v) team assembly and team composition, (vi) undergraduate students and graduate students
• 2016-2017: (i) undergraduate students and learning, (ii) biomedical research and effectiveness, (iii) leadership and team science skills, (iv) promotion and tenure
• 2018-2019: (i) Clinical and Translational Science Award (CTSA) and funding, (ii) gender and productivity, (iii) teaching and curriculum, (iv) team science training and graduate students, (v) promotion and tenure.

We then mapped the results in Figure 3:

• The research topics are illustrated as coloured bars. Where related topics occurred in two consecutive time periods, they are shown using the same colour. Same-coloured bars are connected using grey tubes. The width of the tubes is proportional to the number of titles and abstracts accounted for in the clusters. Darker tubes are more robust in that they share more nodes between two consecutive time periods (Laboratoire Interdisciplinaire Sciences Innovations Sociétés, 2020).
• We also connected related topics using red arrows. This shows that there were five connections among topics from 2010 to 2014, rising to eight connections between 2014 and 2018. In addition, the red arrows show that topics could:
  • continue relatively unchanged (“Continuation” where the topic “impact & evaluation/assessment” is a continuation of “impact & effectiveness”)
  • spilt into different but related topics (“Differentiation” where the topic “impact & evaluation/assessment” evolved into “evaluation/assessment & translational research” and “impact & productivity”)
  • combine with related topics (“Fusion” where “undergraduate students and learning” and “biomedical research and effectiveness” combine into “teaching and curriculum”).
• There were also several topics which only appeared once in the ten-year period, see for example “community and Team Science Toolkit” in the first time slice (2010-2011), which is labelled as an “isolated community.” Other (unlabelled) examples are “team formation and model” in the second time slice (2012-2013), and “gender and productivity” in the fifth time slice (2018- 2019). There were four isolated research topics from 2010 to 2013 and three isolated research topics from 2014 to 2019. These can also be interpreted as new emerging topics.

Figure 3 shows that the most widely investigated topic was “evaluation/assessment & translational research” in the third time slice (2014-2015), but it did not evolve continuously thereafter. This does not mean that research on this topic disappeared, but rather that the topic is not highly correlated with the hot topics in the later time slices. Also interesting from an evolutionary perspective is the change in education-related topics from 2012 onwards. In the second time slice (2012-2013) there were two related topics “team science training and learning” and “undergraduate students and education” and these evolved – through differentiation and fusion – into “teaching and curriculum” and “team science training and graduate students” in the fifth time slice (2018-2019).

Overall, the results indicate that there are less isolated research topics and more connections between topics in recent years.

Conclusion

We have presented methods that can be used to show how research evolves and applied it to the science of team science. We note that the results are not always consistent across methods, for example, while “evaluation and assessment” is the most used term across the 10 years, the topic of “evaluation/assessment and translational research” has not evolved continuously after the third time slice (2014-2015). However, “team science training” not only shows up consistently when analysed as a term, but also shows good evolutionary continuity in the analysis of the evolution of topics.

Are our findings consistent with your experience? Can you suggest reasons for the formation and evolution of research topics that we have shown? Do you have other suggestions for analysing featured topics and tracking their evolution? Are there additional analyses that we should undertake? What do you predict will be the next emerging topics? We are keen to hear your thoughts and comments.

References:
Falk-Krzesinski, H. J., Contractor, N., Fiore, S. M., Hall, K. L., Kane, C., Keyton, J., Klein, J. T., Spring, B., Stokols, D. and Trochim, W. (2011). Mapping a research agenda for the science of team science. Research Evaluation, 20, 2: 145-158.

Laboratoire Interdisciplinaire Sciences Innovations Sociétés. (2020). Cortext Manager Documentation. (Online): https://docs.cortext.net/analysis-mapping-heterogeneous-networks/mapping-dynamical-analysis-options

This blog post is based on a lightning talk presented at the 11^th Annual International Science of Team Science Conference in June 2020, which was a virtual conference. For more on the conference see: Applying human-centered design to virtual conference planning by Kristine Glauber, Ben Miller and Christine Ogilvie Hendren https://i2insights.org/2020/09/15/human-centered-conference-design/.

Biography: Ying Huang PhD is an associate professor at the School of Information Management at Wuhan University in Wuhan, Hubei, China and a research fellow at the Centre for Research and Development Monitoring (ECOOM ) at KU Leuven in Belgium. His research interests include quantitative science studies and science and technology assessment.

Biography: Ruinan Li is a doctoral researcher at the School of Information Management at Wuhan University in Wuhan, Hubei, China. Her research interests include team science and scientific evaluation.

Biography: Yashan Li is a masters student at the School of Public Administration at Hunan University, in Changsha, Hunan, China. Her research interests include science and technology policy and team science.

Biography: Lin Zhang PhD is a professor at the School of Information Management at Wuhan University in Wuhan, Hubei, China. Her research interests include quantitative science studies and research policy.

 

---

 

团队科学中热点主题的演变 / Evolution of hot topics in team science

An English version of this post is available

团队科学学（SciTS）领域有哪些研究热点？它们在过去十年间如何演化？

基于2010-2019届国际团队科学学（INSciTS）年会的会议议程和CorTexT平台（https://www.cortext.net/），我们从852篇论文和报告的标题和摘要中抽取并清洗了使用频率较高的术语（关键词）作为本研究的数据来源。

高频术语及其演变

图1呈现了使用频次最多的前25个术语及其演变情况，术语及词频分别为：评价/评估（89）、团队科学培训（63）、领导力（31）、交流（30）、课程（14）、社区（29）、资助（13）、混合方法（13）、模型（13）、工具箱项目/工具箱对话倡议（13）、临床和转化科学奖（11），研究生（11）、网络（26）、教育（20）、工具（20）、生产力（16）、文献计量学（19）、框架（14）、性别（11）、准备状态（8）、分布式合作（7）、多团队系统（7）、研究网络工具（8）、团队绩效（7）、多层次（5）。

其中，使用频率最高的两个术语分别是“评价/评估”和“团队科学培训”。在这10年中，“评价/评估”一直以高频率出现；而2014年后，“团队科学培训”的使用频率有显著提升。此外，从图1还可以观察到其他术语使用频率的变化趋势。

研究主题及其演变

基于社区探测算法（Louvain Algorithm），可以将这些术语归纳为6个“社区”，参照Falk-Krzesinski等（2011）进行团队科学学概念映射研究时提出的主题名称，可以将其解释为5个主题（如图2所示）。

1. 团队科学的测量与评价；
2. 团队的机构支持和专业发展；
3. 团队的特征和动态；
4. 团队的管理和组织；
5. 团队的结构和环境。

可以看到，“团队的特征和动态”形成了两个独立的社区，这可能是由于数据量小造成的。在我们的研究中，Falk-Krzesinski等人确定的其他2个主题（“团队科学的定义和模型”和“学科动态与团队科学”）并没有形成独立的社区，而是与其他5个主题相互交织在一起。

我们进一步分析了这些研究主题的演变情况，具体如下：

首先，我们使用上文中提到的社区探测算法（Louvain Algorithm）来确定十年间每2年的重点研究主题（将10年数据划分为5个时间片），并将社区中共现频率最高的2个术语作为该社区的主题标签（不是直接使用图2中使用的标签）。最终得到了23个主题，每一阶段的主题分布如下：

• 2010-2011：1）社区&团队科学工具包；2）社会网络分析&社会网络；3）影响&效能。
• 2012-2013：1）团队组建&模型；2）晋升&终身教职；3）影响&评价/评估；4）团队科学培训&学习；5）本科生&教育。
• 2014-2015：1）成果/产出&测量；2）评价/评估&转化研究；3）影响&生产力；4）团队科学技能&学习；5）团队组配&团队组成；6）本科生和研究生。
• 2016-2017：1）本科生&学习；2）生物医学研究&效能；3）领导力&团队科学技能；4）晋升&终身教职。
• 2018-2019年：1）临床和转化科学奖（CTSA）&资助；2）性别&生产力；3）教学&课程；4）团队科学培训&研究生；5）晋升&终身教职。

这些主题的演化情况如图3中所示：

• 研究主题以彩色的条形图来进行展示。如果在两个连续的时间段内，有2个以上的主题相互关联，则以相同的颜色出现。相同颜色的条形图用灰管连接。管的宽度与聚类中涉及到的文献（会议论文和报告）记录数量成正比。管的颜色越深表明越稳健，因为它们在两个连续的时间段内共享了更多节点（Laboratoire Interdisciplinaire Sciences Innovations Sociétés，2020）。
• 进一步使用红色箭头来连接相互关联的主题。在2010-2014年期间，研究主题之间产生了5个连接，而2014-2018年期间则上升至8个连接。此外，红色箭头还显示，这些研究主题的演变呈现出以下特征：
  • 延续，研究主题演化成相关主题（图中“Continuation”，其中，“影响&评价/评估”主题是“影响&效能”的延续）。
  • 分化，研究主题分化成不同但相关的主题（图中“Differentiation”，其中，“影响&评价/评估”主题演变为“评价/评估&转化研究”和“影响&生产力”）。
  • 融合，相关主题之间进行融合（图中“Fusion”，其中，“本科生&学习”和“生物医学研究&效能”融合为“教学&课程”主题）。
• 还有一些主题在十年间只出现过一次，例如，第1个时间片（2010-2011年）中的“社区&团队科学工具包”可以被认为是“孤立的社区”。类似的例子还有第2个时间片（2012-2013年）中的“团队组建&模型”，第5个时间片（2018-2019年）中的“性别&生产力”。在2010-2013年中，共有4个孤立的社区，而2014-2019年中仅有3个孤立的社区。这些孤立的主题也可以理解为新出现的新兴主题。

值得注意的是，在第3个时间片（2014-2015年）中，最大的研究主题是“评价/评估&转化研究”，但此后该主题并没有发生进一步的演化。这并不意味着关于这个主题的研究消失了，而是表明这个主题与后一个时间片中热点主题的相关性不高。从演化的角度来看，同样有趣的是，在2012年以后，与教育相关的主题变化。在第2个时间片（2012-2013年）中，有两个与教育相关的主题，“团队科学训练&学习”和“本科生&教育”，这两个主题通过分化和融合，在第5个时间片（2018-2019年）中演变为“教学&课程”主题和“团队科学培训&研究生”主题。

总的来说，分析结果表明，近年来孤立的研究主题较少，主题之间的联系越来越紧密，主题演化呈现出较好的连续性。

结论

在本研究中，我们展示了可用以分析研究主题演变的方法，并将其应用到了团队科学学的研究主题分析中。我们注意到，不同方法的结果并不总是一致的，例如，虽然“评价/评估”是10年中被使用最多的术语，但“评价/评估&转化研究”这一主题在第3个时间片（2014-2015年）之后并没有发生进一步的演化。然而，“团队科学培训”术语不仅被使用的频率很高，同时在主题演化分析中，也显示出较好的演化连续性。

我们的分析结果与您的研究经验是否有相似之处？您能否就我们所展示的研究主题的形成和演变提供一些可能的原因解释？您对于关键主题识别与追踪是否有进一步意见和建议呢？我们还可以（应该）进行哪些其他的分析呢？您预测团队科学学研究中下一个新兴主题会是什么呢？我们很期待听到您的想法和意见。

参考文献

Falk-Krzesinski, H. J., Contractor, N., Fiore, S. M., Hall, K. L., Kane, C., Keyton, J., Klein, J. T., Spring, B., Stokols, D. and Trochim, W. (2011). Mapping a research agenda for the science of team science. Research Evaluation, 20, 2: 145-158.

Laboratoire Interdisciplinaire Sciences Innovations Sociétés. (2020). Cortext Manager Documentation. (online): https://docs.cortext.net/analysis-mapping-heterogeneous-networks/mapping-dynamical-analysis-options

这篇博客文章是基于第11届年度国际团队科学学会议中的一个闪电演讲形成的（该会议是一个虚拟会议）。https://www.inscits.org/2020-scits-conference.

个人简介

黄颖，武汉大学信息管理学院副教授，鲁汶大学研究与发展监测中心（ECOOM）研究员。研究兴趣为定量科学研究、科技评估。

李瑞婻，武汉大学信息管理学院博士生，研究兴趣为团队科学、科研评估。

李雅珊，湖南大学公共管理学院硕士研究生，研究兴趣为科技政策、团队科学。

张琳，武汉大学信息管理学院教授，研究兴趣为定量科学研究、研究政策。