Diana Kwon in Nature:
Underground creepy crawly state. Bosom malignancy. Sun oriented force. These might sound like expressions from a work of fiction, but they are actually strange translations, pulled from the scholarly literature, of scientific terms — ant colony, breast cancer and solar energy, respectively. Guillaume Cabanac, a computer scientist at the University of Toulouse, France, spots such bizarre phrases in academic papers every day.
This year, Cabanac and his colleagues found these tortured phrases, as they call them, in thousands of papers. A handful have been retracted; publishers are investigating many more. Cabanac has built a website to keep track of the mushrooming problem. “They found this whole new hornet’s nest of articles that appear to be completely fake,” says Elisabeth Bik, a research-integrity analyst in California. Weeding out these problems is related to Cabanac’s day job: he specializes in analysing the scholarly literature, and now devotes around two hours a day to finding tortured phrases. Some people might find them funny, but Cabanac takes the problem seriously. “This shouldn’t be happening,” he says.
Cabanac’s hunt for gibberish papers began in 2015, when he started collaborating with Cyril Labbé, a computer scientist at the University of Grenoble Alpes in France. Labbé had developed a program to spot gibberish computer-science papers automatically generated using SCIgen, a piece of software created initially as a joke. Labbé’s work led journals to withdraw more than 120 manuscripts. Cabanac helped to update Labbé’s program to find papers only partially written by SCIgen, and to locate them using Dimensions, a search engine for scholarly literature. This year, they reported finding hundreds more papers containing nonsense text, published in journals and conference proceedings and as preprints.