Skip to main content
EU Science Hub
  • News announcement
  • 10 June 2020
  • 4 min read

JRC to release AI tech for coronavirus fact-checkers

An innovative open-source AI text mining software, available for the fact-checking community, could help fight against coronavirus disinformation.

The JRC’s tools will help tackle the wave of false or misleading information that has come during the coronavirus pandemic
The JRC’s tools will help tackle the wave of false or misleading information that has come during the coronavirus pandemic
© ©EU 2020

In the fight against coronavirus disinformation, the JRC has announced plans to make innovative AI text mining software open-source and available for the fact-checking community.

The JRC-developed tools use AI technology to automatically identify and monitor disinformation, from flagging fake news to tracking dodgy sources. By doing so they can help to keep the facts straight on coronavirus, which is essential to an effective public health response.

Tackling disinformation to protect our health

The coronavirus pandemic has brought with it an unprecedented ‘infodemic’. A flood of information about the virus is available and shared online, and a lot of it is false or inaccurate. This risks creating confusion and distrust, and ultimately being dangerous to people’s health.

In response, the Commission and the High Representative have released a Joint Communication setting out key challenges faced in the current crisis and concrete actions that can be quickly set in motion to tackle them.

One such action is a commitment to step up support to fact-checking and research activities by providing tools to detect, analyse and expose disinformation. The JRC is now working to open-source two key tools it has, so the community fighting coronavirus disinformation can use and adapt them.

Using AI to detect fake news – the Misinfo Classifier

Most fake news is written with language chosen to deliberately arouse heightened emotions of fear and disgust. This means that experts can train software programmes to start to identify patterns in the language and quickly identify whether something might be fake news or not.

The Misinfo Classifier detects fake news by measuring the ‘shrillness’ of language in a news article, using a machine-learning algorithm developed by the JRC. There is no such thing as a 100% detection rate for fake news, but the JRC’s machine learning algorithm has about an 80% hit rate. This is comparable to the state of the art right now.

The tool is currently being used by the Commission and European Parliament and will be made available to reputable fact-checking organisations in the coming weeks, so that experts can more easily identify possible fake news and debunk the claims being made.

Tracking dodgy messages and who is sending them – the Social Rumours tool

Social Rumours is a kind of ‘treasure hunt’ tool for dodgy websites, helping experts to quickly get a good picture of where fake stories are coming from. The software identifies a number of accounts on Twitter that are reposting links to a known ‘unverified’ or dodgy source and looks to see what other links they are posting to get an idea of the kinds of narratives that are emerging on social media.

The tool is also being used by the Commission and European Parliament and will be made available to the same organisations who wish to access the Misinfo Classifier tool, so that they can better identify new and emerging disinformation narratives.

Working with others for a global response

JRC experts have a long-standing collaboration with the World Health Organization, which has proven vital in tackling this pandemic effectively and will be put to good use in making the actions set out in the Commission communication a success. The Epidemics Intelligence from Open Sources (EIOS) platform is a collaboration between the WHO and the JRC based on the JRC’s Europe Media Monitor Medical Information System (MEDISYS) tool that is collating up to 120,000 articles per day related to the coronavirus pandemic.

The platform is helping to sort through this information and make it available to experts across the globe who are tracking the pandemic. It helped WHO to detect the first signs of the COVID-19 outbreak in Wuhan at the end of December last year, and has since been used to trace the spread of the disease globally – up to and including when the disease arrived on our doorstep.

Data from EIOS also feeds in to the COVID-19 News Map, which displays the ten most recent articles related to COVID-19 in their original languages.

Background

The exposure of citizens to large scale disinformation, including misleading or outright false information, is a major challenge for Europe. The Commission is working to implement a clear, comprehensive and broad set of actions to tackle the spread and impact of online disinformation in Europe and ensure the protection of European values and democratic systems.

The JRC carries out extensive research on fake news. From identifying sources of disinformation to tracking trending narratives, this work seeks to detect such stories early, and provide analysts and spokespersons with the tools to counter dangerous narratives.

The Europe Media Monitor (EMM) developed by JRC covers about 300,000 articles per day from traditional news sites. With such an overwhelming volume of information, specific software is necessary in order to classify the information by policy area, and to cluster it in order to find the top stories.

Related Content

Coronavirus: EU strengthens action to tackle disinformation

The digital transformation of news media and the rise of online disinformation

Daily COVID-19 media surveillance

Have your say on the effects of COVID-19

Details

Publication date
10 June 2020