How publishers are fighting fraud

According to the National Science Foundation, peer-reviewed science and engineering journal articles and conference papers have grown by about 4% annually over the past 10 years. Challenges driven by net zero commitments, for instance, are creating ever-greater interest in research on clean and green technologies, accelerating the growth of published papers on these topics.

To be able to confidently study these subjects, and depend on the knowledge that they have gleaned from articles and papers, engineers must be able to trust in the reliability of published research. Unfortunately, there is now a significant rise in the number of fake papers being published (see The cost of fraudulent research).

In response to this issue, Elsevier has significantly expanded its Research Integrity and Ethics team, growing it from two to more than 20 people in just a few years.

“Misconduct is so much more rife than it used to be,” says Dr Oliva Nippe, a Senior Publishing Ethics Expert on the team. “The paper mill industry has boomed, and it looks as though these papers have been appearing and amassing since about 2017. So there are just more bad papers out there than there used to be.”

“It's really important to shift from a reactive stance to a proactive stance, to be able to identify issues pre-acceptance — that's key,” says Dr Daniel Stuckey, also a Senior Publishing Ethics Expert on the team.

A range of methods are used to get ahead of the problem, from providing education and guidelines to authors and editors, to using advanced software and AI tools.

Dr Stuckey highlighted iThenticate, which combs a massive database for evidence of text similarity between published papers. “That can be also really useful to identify suspected plagiarism,” he says. “There is the duplicate submission check, which can look for duplicated submissions across different Elsevier journals,” he adds, “and there's also a paper mill detection tool that’s been developed in-house, and we're currently testing that out.”

These tools are meant to assist editors by alerting them to potential issues — they don’t tell you if a paper does or does not have a fraud problem. It’s never that simple. But they can at least provide an investigative lead.

For instance, the Identity Confidence Check (ICC) is a critical tool developed by Elsevier’s Trust and Transparency team: Elsevier’s editors use it to assess the identity of an author when they first submit a paper for consideration. It takes various factors into consideration. As Dr Nippe explains: “It'll look at the [author’s] email domain. Is it related to an institution? Is it a domain that is associated with misconduct in the past? Is the author on the editorial board for that journal? Have they authored papers at that journal before? Have they been a reviewer? All things that might establish more confidence that they are the person they say they are, and that they do have expertise in the field.”

Depending on how many of those flags are checked and how many pass muster, the author will get a green, amber or a red icon (or gray if there isn’t enough information available). “We have some resources that we’ve given to the editors so they can understand what those flags mean and interpret them accordingly,” says Dr Nippe, noting that even a red label doesn’t mean they’re necessarily a “bad actor“ who is trying to submit something fraudulent.

“It just means that maybe the editor should keep an eye open for any potential things that might be wrong with the paper,” she says. “Just be a little bit more cautious."

In the worst-case scenario that a paper with faulty research gets past the review process, it’s vital to be able to identify and correct it as soon as possible. This will typically happen in response to new information supplied by an editor or reviewer. When that information is presented, Elsevier immediately launches a formal investigation.

Ultimately, based on the severity of the problem, there are steps Elsevier can take. In some instances where there was a relatively minor error that can be fixed, the article may be updated with a corrigendum (i.e., a correction) or an editorial note (if the authors do not agree with the text) explaining what changes were made. If there is an actual legal issue, it’s possible the paper could be removed entirely.

When there is an issue of fraud or an ethical breach, that usually leads to a retraction. This could be the result of the study being unreliable because of falsified data, a compromised review process, or various other causes. The paper will be formally retracted, including the placing of a large watermark and a link to the retraction notice explaining why the paper is not reliable.

Whichever of the available options is chosen, the final decision on how to appropriately correct the record always rests with the journal’s Editor-in-Chief.

In the case of retraction, just retracting the paper isn’t enough, however. Once that research has been released into the world, there is the risk that other researchers will unknowingly use or reference it, effectively promoting false information or data. That’s where the RetractoBot comes in. Utilizing Scopus citation data, the University of Oxford’s Bennett Institute for Applied Data Science runs RetractoBot as a service to alert researchers if they have cited a retracted paper. This is one important way of reducing new citations to articles that have been retracted.

“It's always keeping in mind that other people must have reliable data that they can go back to,” says Dr Nippe of the retraction process. “Ultimately, it's about preserving the scientific record.”

Artificial intelligence presents both opportunities and challenges when it comes to ethics in publishing. AI can be a great help to researchers who need assistance in organizing their thoughts or clearly stating important information in the text of a paper. Also, AI has been used successfully by editors to detect duplicate images. But AI also has the potential to create false images. Meanwhile, catching completely new graphics or images generated by AI, perhaps based on fake research data, is a different challenge entirely.

“This is something that the publishing industry will have to look at, and there are possible solutions,” Dr Stuckey notes. “For example, using digital watermarks that will make it harder to be able to pass off generative AI created images as your own within a manuscript.” But there is no question that this will be a complicated problem to tackle.

At Elsevier, we hope our rigorous processes and methods help researchers feel confident that they can trust our publications. The data seems to bear out that they do.

While 18% of articles were published by Elsevier from 2017 to 2021, during that time we had a 27% share as measured by citations, which indicates that leaders in research trust our publications.

Furthermore, we’re honored to work with the authors and editors that help us deliver an even greater percentage of the world’s highest quality research as measured by Field Weighted Citation Impact (FWCI):

• 33% of the top 10% journal FWCI tier
• 29% of the 10-25% journal FWCI tier

We hope to be able to continue and grow our reputation as a leading publisher spotlighting top research from around the globe.

While no publisher can guarantee that fraudulent papers or falsified information won’t ever make it to publication, we work to build trust with our readers and partners by maintaining high standards, constantly seeking new ways to improve our processes, and investing in new tools. This includes involvement with initiatives and organizations like the STM Integrity Hub, COPE and the European Association of Science Editors.

“We're getting there,” Nippe adds. “We're all developing our tools. We’re all expanding our teams. We're all increasing our knowledge,” she says, noting that it’s always an “arms race” between the ethics teams and the fraudsters.

But she also emphasizes how much Elsevier strives to bring trust and value to authors, the vast majority of whom are dedicated researchers who are eager to share their work with a broader audience. “We put a lot in to make sure that we have credible, trusted titles and that we support the authors as best as we can on their submission journey.”

With the help of the Research Integrity and Ethics team, Elsevier is proud to be building a body of published research that you can trust.

Read the full article, Fighting the Problem of Fraud in Publishing, to learn more about Elsevier’s efforts to ensure integrity in publishing.

#MIsinformation #FraudDetection #Research