- Used VOSviewer to map 2,115 most relevant terms in titles and abstracts that were mentioned in at least 130 publications
- Clear distinction between biologically and clinically focused topics 
- There is more of an emphasis on clinically focused topics and health research in recent years

A scientometric overview of CORD-19 data analysis - term map

- Looked at the topics of 140,302 articles, as defined by the probability distribution over a vocabulary of publication titles and abstracts 
- Latent Dirichlet Allocation (LDA) model, determined 15 topics worth observing 
- Literature published in 2020 mostly focused on COVID-19, its outbreak, management, and consequences 
- Concluded that coronavirus research becomes more popular after an outbreak
- Used agglomerative clustering to further divide up topics and discovered that CORD-19 contains mostly publications on coronaviruses compared to other literature on health, epidemics, and clinical medicine in general 
- How active CORD-19 is largely depends on the time and outbreak 
- COVID-19 outbreak has generated far more literature compared to past coronavirus outbreaks 

A scientometric overview of CORD-19 data analysis - topic modeling

- Looked at references to papers in CORD-19 (“external” publications) and references in CORD-19 papers
- 1,844,831 publications and 3,974,851 citation links, of which 104,508 publications are in CORD-19 
- Used Leiden algorithm and assigned publications in CORD-19 a node weight of 1 and “external” publications given a weight of 0
- 11 clusters, each containing more than 2,000 CORD-19 publications (accounting for 54% of CORD-19 publications, the largest group)
- Added the citations cluster to term map used earlier 
- The largest cluster was made up of 16,913 publications on COVID-19 

A scientometric overview of CORD-19 data analysis - citation network analysis

- 88,570 publications in CORD-19 (63%) have been mentioned in Altmetric, a much higher percentage than any other study on publications in specific fields
- Much higher mention (68%) of recent publications (published in early 2020) in social media as described by Altmetric 
- Twitter accounts for the majority of social media interactions studied – approximately 87% of Twitter mentions are regarding publications from 2020
- Twitter mentions also correlated with the most recent publications 

A scientometric overview of CORD-19 data analysis - altmetric analysis

- Focused on 140,302 out of the 169,821 available publications on CORD-19 collected on July 1, 2020, disregarding articles that did not have an DOI, PubMed ID, PubMed Central ID, or arXiv ID 
- Term map
- Topic modeling
- Citation network analysis
- Altmetric analysis

University of Michigan - Ann Arbor

- Analyzed COVID-19 and coronavirus research publications in the CORD-19 database
- The dataset includes publications beyond COVID-19 and coronavirus research, but compared to Web of Science, it is evident that CORD-19 still contains the majority of COVID-19 and coronavirus-related literature 
- Topic modeling demonstrated that research follows specific virus outbreaks and have temporal patterns 
- Citation network clustering showed that there are two significant citation clusters – one that is mainly focused on COVID-19 and another on molecular biology 
- Looking at Altmetric data, it can be concluded that Twitter is the most active social media in terms of mentioning CORD-19 publications, especially regarding most recent literature on COVID-19 

A scientometric overview of CORD-19

Colavizza, G., Costas, R., Traag, V. A., Van Eck, N. J., Van Leeuwen, T., & Waltman, L. (2021). A scientometric overview of CORD-19. PloS one, 16(1), e0244839. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0244839 

- Dataset containing COVID-19 publications that is updated every day 
- Dataset released to the public to allow researchers to add findings to existing research
- A collaboration with academic institutions and is backed by the US White House

COVID-19 Open Research Dataset (CORD-19)

- Used the July 1, 2020 version of CORD-19 that contained 169,821 articles 
- Publications were collected from Medline and PubMed’s PMC open access corpus, WHO database of COVID-19 research articles, and preprints from arXiv, bioRxiv, and medRxiv
- Used Dimensions to study the number of citations and fields of research on publications in CORD-19
- Used WoS data for looking at whether relevant COVID-19 publications are left out of CORD-19

A scientometric overview of CORD-19 data collection

A scientometric overview of CORD-19 data analysis

- Continued mapping of COVID-19-related research 
- Contributing datasets on COVID-19 research to CORD-19 to take advantage of open data sharing
- Online scientific information diffusion through altmetrics or other data sources

Learn Before

Related

Learn After