˾ֱ

Using AI, ˾ֱ researchers develop smart search tool to help navigate flood of COVID-19 work

A researcher stands over a lab bench
More than 2,000 pre-prints of COVID-19 studies have appeared since January, making it difficult for researchers to keep tabs on the coronavirus-related work that's taking place around the world (photo by Rober Solsona/Europa Press via Getty Images)

Scientists around the world have been racing to respond to the deadly coronavirus pandemic. Seemingly overnight, many of the world’s top researchers shifted their focus to the study of COVID-19, leading to a constant stream of new pre-print articles.

More than 2,000 COVID-19-referencing pre-prints – completed studies awaiting peer review and journal publication – have appeared since China announced the outbreak in January, with 375 articles published last week alone.

Now, researchers at the University of Toronto have developed a tool, called CiteNet, to help their colleagues find, survey and review this new literature in a faster more efficient way.

“We wanted to come up with something more intuitive that allows you to explore literature and see what’s out there,” says Duncan Forster (left), a graduate student co-supervised by Charlie Boone and Gary Bader, both professors of molecular genetics in the Faculty of Medicine and at the Donnelly Centre for Cellular and Biomolecular Research.

 

The trio co-developed CiteNet with fellow graduate student John Giorgi in collaboration with Bo Wang, an assistant professor in the department of medical biophysics in the Faculty of Medicine and a faculty member at the Vector Institute for Artificial Intelligence.

“If you find a paper that you think is interesting, CiteNet can help you find others like it,” says Forster.

CiteNet indexes papers from the pre-print servers BioRxiv and MedRxiv (pronounced as “bio-archive” and “med-archive”), where most COVID-19 papers appear before publication. But instead of the needle-in-a-haystack approach of keyword searches employed by most academic search engines, CiteNet uses algorithms to intelligently gather literature related to COVID-19 and sort it based on defined search criteria.

“Using advances in natural language processing, CiteNet scans papers for semantic similarities and ranks them based on their likely relevance to the query papers,” says Wang (left), who is also lead artificial intelligence scientist at the Peter Munk Cardiac Centre and the Techna Institute at the University Health Network.

 

CiteNet is still in the development phase, but due to the COVID-19 pandemic, Forster and Giorgi have made the demonstration version of the tool available to the public. They have also created and posted a CiteNet video tutorial to illustrate how to use the app.

“We hope CiteNet will be a useful, up-to-date tool for all members of the scientific community looking to find answers in the global fight against COVID-19,” Wang says.

Donnelly