How ClaimReview is simplifying the process of fact-checking

Behind the scenes, Interviews

One of the biggest challenges facing journalism today is the prominence and spread of unverified facts and misinformation. In an era marked by fake news, it has become increasingly difficult to distinguish the accurate from the inaccurate. That is why Joel Luther and the team at the Duke Reporters’ Lab collaborated with Google to create ClaimReview, a simplified way of logging fact-checks so that readers have easy access to a global database of verified claims. 

This tool, now used by more than half of the world’s fact-checking platforms, summarizes a fact-check and identifies key information allowing platforms like Google and Bing to highlight and display important facts to their readers.

Luther spoke with Storybench to discuss how ClaimReview works and why fact-checking is important in today’s media landscape.

What is ClaimReview?

ClaimReview is a tagging system that writers of fact-checks can use to signify that their articles are fact-checks and to highlight some of the key information. Some of the key pieces it will include are what the claim was, who the speaker was, what the rating was, and a conclusion on the accuracy of the claim. That information then gets translated into a machine-readable format so that it can be surfaced by Google, Bing, or any other platform that wants to use it however they want to use it. 

How does the process work?

So if you are a writer for Politifact, the Washington Post, or any other fact-checker, the easiest way to use it is through a tool that we have developed with Google called the Fact Check Markup Tool. Essentially it’s just a form. You log on after you’ve written your fact-check, and it will ask you some basic questions. Then you just hit submit and it will go off into an accessible data feed.

How did the idea for the project come about?

It actually came from a suggestion by Glenn Kessler, who is the chief writer of The Fact Checker, the fact-check column at the Washington Post. He said, “Wouldn’t it be cool if there was a way for search engines to be able to highlight fact-checks?” So that started the process with the IFCN (The International Fact-Checking Network) and then also with platforms like Google, Bing and Jigsaw. That also led us to the community, which is a tagging system used for any type of data, but we worked with them to make it specifically for fact-checks. So it was an open process and people could give their input, opinions, and thoughts on how it should function. Then all that was synthesized into what we have now. 

What has your role been on the project?

This is the second iteration of ClaimReview. It started several years ago and at the time publishers had to embed javascript into their website when they were publishing a fact-check. The Fact Check Markup Tool from Google is fairly new. My job has been to get people switching over to the new way of using ClaimReview which is, we think, easier. I’ve also been responsible for outreach to fact-checkers globally to encourage them to use it as well. 

What are the criteria for distinguishing a fact-checker?

We have certain criteria that we use. We consider a fact-checker to be any media outlet that regularly publishes articles that assess the accuracy of a statement made by a candidate or political figure. One of the biggest ones is Politifact. However, we also include the Washington Post and the New York Times because they both have fact-checking columns or sections. It’s a growing format that started in the mid-2000s. I think was the first major one, and it has expanded to over 200 now. 

You mentioned you are hoping to see ClaimReview used by more fact-checkers globally. How are you planning to do that?

We do a census of fact-checkers here at the Duke Reporter’s Lab, so we have a database of every fact-checker we have identified in the world. That number is currently at 226 and the number of people who have used or currently are using ClaimReview is at 118. So at the moment, we have a little over half of the world’s fact-checkers using our tool. Ideally, I think we would see every fact-checker using it. The bulk of my job is direct outreach to fact-checkers who we’ve identified in our database as fact-checking outlets who aren’t yet using it and informing them on why this is important and how they can get started using it.

What are some of the innovative or unique ways this tool is being used?

I would say that the innovation behind its use comes down to a platform by platform decision as opposed to a journalist’s decision. It’s a pretty standardized way of tagging. Any time you publish one you can say, “Here is the claim, and here’s how I rated it.” But then that gets sent publicly to anyone who wants it. So there are a couple of cool ways this is being used. Google will surface a fact-check that has been marked up with ClaimReview in search results, and it will highlight it as a fact-check. So if I search, “Did Donald Trump oppose the war in Iraq?” one of the first results that will be on the page will be a fact-check, and it will say what the claim is, who made the claim, and the rating of that claim, so you can easily see all of that in the search results. Bing does something similar. Google News has a dedicated section on the Google News homepage that lists the latest fact-checks. 

At the Reporter’s Lab, we are building a couple of cool tools to use it. Fact Stream is our mobile app that regularly updates with fact-checks as they come in and as they are tagged with ClaimReview. We are also building an on-screen TV fact-checking experience called Squash. We have this huge database, I think it’s over 40,000 fact-checks that have been tagged with ClaimReview, and we can search through that anytime a politician says something and display a fact-check on the screen if it matches one that we find.

What is your vision for the future of ClaimReview?

One of the big things that we are doing now is working on a follow-up to ClaimReview that we are calling MediaReview. That would be a version of ClaimReview that could be used for fact-checks of images and videos. We see frequently that manipulated images and videos are going to be a massive problem in the future, and they’re already becoming a problem. So we’ve been developing a new tool that will enable fact-checkers to tag a video as being edited, transformed, or doctored in some way. We hope to get that into wide adoption and also think that can help with a consistency of language around fact-checks of those types of media. It’s kind of mix-matched and all over the place now in terms of what words people are using to describe the same thing.

Why is ClaimReview, and fact-checking in general, important in today’s media landscape?

Fact-checking is a vital field of journalism for keeping politicians accountable and keeping the citizenry informed. ClaimReview expands the reach of such an important form of journalism and is the backbone of cutting-edge tools being used to show more fact-checks to more readers in a timely fashion. It’s the secret sauce that will enable us to realize the dream of automated fact-checking.

Latest posts by Kate Secrest (see all)

Leave a Reply