Artificial intelligence scrutinizes controversial science publications, aided by human intervention

Article Title: New AI Model Aims to Combat Predatory Journals in Open Access Publishing

In the world of scientific research, concerns about predatory and hijacked journals have been a persistent issue for over a decade. These questionable publications, which often exist primarily to extract fees from academics, have been a thorn in the side of the scientific community.

A team of three computer scientists from the University of Bonn Institute of Computer Science, the Faculty of Computer Science at the University of Saarland, and the Faculty of Computer Science at TU Dortmund University, have developed a machine learning classifier to help identify these problematic journals.

The researchers analysed a data set of nearly 200,000 open access journals, settling on 15,191 for a closer look. Their model flagged 1,437 questionable journals, but missed about a quarter of the time. Out of these flagged journals, approximately 1,092 are expected to be genuinely questionable, while about 345 are false positives (24% of the flagged set).

To validate the model's results, the researchers conducted a human review of the flagged journals. They identified questionable journals as those that violate best practices, have low editorial standards, and coax academics into paying high fees for publication in a journal with insufficient editorial review.

The researchers aim to collaborate with indexing services and reputable publishers to help them identify and address these questionable journals. They suggest that by adopting a more stringent setting, only five false alarms out of 240 would be expected.

Open access scientific publishing, which began in the 1990s, aimed to expand the availability of academic research by changing the payment model from institutional subscriptions to author payments. The movement has seen significant growth, with a 2022 memorandum from the White House Office of Science and Technology Policy directing US agencies to make taxpayer-supported research publicly available by the end of 2025.

However, the rise of open access publishing has also led to an increase in predatory journals. These scam journals can change their names and websites with ease, making list-based approaches less effective.

In an effort to combat this issue, Jeffrey Beall, a librarian at the University of Colorado, coined the term "predatory publishing" in 2009. Now, with the development of AI models like the one created by these researchers, there may be hope for a more effective solution.

Acuña, the founder of ReviewerZero AI, an AI service aimed at detecting research integrity problems, may make this tool available to scientists before they submit to a journal in the near future. The research paper detailing this work was published in Science Advances.

Despite the progress made, the researchers are not yet ready to name and shame the dubious journals due to potential legal challenges. However, their work is a significant step towards ensuring the integrity and credibility of scientific research in the open access era.

Artificial intelligence scrutinizes controversial science publications, aided by human intervention