Wilfrid Laurier students create COVID-19 fake news detector
Tool is triggered by phrases like 'manmade virus' or 'drink bleach'
A group of students at Wilfrid Laurier University have come up with a COVID-19 fake news detector to help fight the spread of misinformation online.
The group spent four months developing the analytical tool that determines the probability of a fake news story or article that contains factually-inaccurate information about COVID-19.
"COVID has brought on the world's first 'infodemic'… defined by the World Health Organization. [It] means both true news and fake news spread together and people have a very hard time finding trustworthy sources, and believing in fake news can result in very serious consequences and even costing lives," said Irene Zhang, one of four student researchers on the project and a mathematics master's graduate from Laurier.
"Our goal is simple. We just want to make some contributions toward this misinformation war," she told CBC News.
The tool was submitted in a case study competition by the Canadian Statistical Sciences Institute last month. It is not currently available for public use.
How it works
The group used a "machine learning algorithm" that feeds on input data to generate predictions.
"You can think of these as algorithms that learn from information that's given to them, similar to how we learn, said student researcher Rini Perencsik, who is also a second-year data science student at Laurier. "So in order to create this model, to predict COVID-19 news is false, it needs to learn about COVID and fake news."
"It does this through data in the form of past news articles related to COVID. We retrieve data from online sources and after doing analysis on the data and trying different algorithms, we combined many different models to form one final one," she said.
About 5,000 pieces of information were input to create the algorithm. It's triggered by certain phrases or keywords such as "manmade virus," or "drink bleach," which are often associated with factually inaccurate news articles, Zhang said.
It was initially difficult for the group to find reliable sources, but then the students discovered existing data sets by a collection of researchers, labelled with false and accurate information. That was combined with their own research.
Future of the tool
The group is awaiting competition results to help determine next steps for the tool.
Perencsik said she hopes the algorithm can be a launch pad for future research for institutions with larger data sets. The hope is to continue spreading awareness about false information.
"Many aren't taking the pandemic seriously or are being given false information on how to protect themselves or the virus or just our current understanding of the virus. That's something that's very harmful to society," she said.
With files from CBC K-W's The Morning Edition