French researchers discover 243 nonsense letters

KDo you name the mathematical treatise “On the Continuity of Morphisms”? It is eleven pages long, contains an impressive number of aphorisms and ends with an exhaustive bibliography. The author of these lines, along with his twelve-year-old son, is named as the author. This did not require research. there is one online generator, which produces grand-sounding but meaningless essays on mathematics. All you have to do is enter the name of the desired author.

French researchers Cyril Labe and Guillaume Cabanaque have now identified 243 nonsensical articles in scientific publications. To do this, they searched academic search engine dimensions for grammatical patterns and stylistic formulations that are particularly common in such publications, but rarely by human authors. The discovery of these fingerprints, such as “indeed, some futurists would disagree”, is done using information technology. The results were checked by hand.

Funny or disappointing?

Tools to generate false scientific text are also available for computer science (SCIGen), physics (Sci-Physics) and philosophy (Postmodernism Generator), there is even a subsidy generator (Automatic SBIR Proposal Generator). “People register their names and submit papers to conferences and magazines,” Labey and Cabnack explain to the newspaper. “With this reservation, problematic publications tend to come from certain places in the world.” Of the counterfeiters, 156 came from China, 54 from India, three from Indonesia and one each from Belgium, Iran, Poland, Slovakia and the United States. State. The country of origin cannot be determined for other contributions.

Some papers were also edited by editors to match the style of the academic journal. “Only publishers know the reasons why these papers pass peer review,” say Lebe and Cabnack. You give three possible explanations: Since the English looks correct, a reviewer may be influenced by the jargon. The other possibility is that the title and subtitle are read only superficially. Finally, it is also conceivable that a peer review process is completely missing. But this isn’t the only point where publishers fail. It also takes a long time in many cases to retract nonsensical essays, and publishing houses often say: “We are still investigating.”

Nonsense essays are a problem in science. Two French suspects that only 75 out of a million letters were created by computers. While the existence of these essays is depressing for Cabanac, Lebe finds them quite funny. More frightening of the two are “research paper mills”, i.e. organizations that produce and sell counterfeit manuscripts that give the appearance of genuine research. Falsification of data is also dangerous. So, two of them, together with Jennifer Byrne from the University of Sydney, are working on tools to automatically detect errors in cancer studies (Seek and Blast) Together with Alexander Magzinov, he also tracked down formulations that indicate programs that hide plagiarism. Ultimately, none of these tools could solve the fundamental problem of fraudulent behavior. Incidentally, the treatise “On the Continuity of Morphisms” remained unpublished.