Journals test software to catch cheats

Anti-plagiarism software similar to that used to catch cheating students is to be turned on academics by journal publishers.

Universities across the UK monitor students' work using Turnitin, a program that assesses the origin-ality of assignments.

Now a new program based on the same technology is being developed to apply similar high-tech scrutiny to research papers.

CrossCheck was designed by iParadigms, the creator of Turnitin, in collaboration with the publishing association CrossRef.

It has already been tested in a pilot project that involved eight journal publishers, including three based in the UK.

The software, which is due to be released in June, compares manuscripts against databases of millions of articles and produces a report listing any material that overlaps.

Phil Caisley, head of information services at the BMJ Group, said the software could have two applications - checking manuscripts before they are published to see if a submitted article uses other work without attribution and checking after publication "to see who is ripping our content off".

"During the pilot, we found much more activity in post-publication use of content than we did in pre-publication plagiarism," he said.

Trish Groves, deputy editor of the BMJ, said it was impossible to know how widespread plagiarism was.

"We do hear when it happens, but in a very ad hoc way. We publish something and then we get e-mails saying, 'Hang on a minute, that's my article,'" she said.

"It's a pretty rare event, once a year perhaps, but ... it's only when we start checking every article prior to publication that we'll know how much of it is out there."

In trials of CrossCheck involving four BMJ Group journals, only one article out of the hundreds that were screened was flagged up as possible plagiarism.

However, the BMJ alone receives 8,000 submissions a year, and the BMJ Group has 23 other journals.

Dr Groves said publishers may consider targeting authors judged most likely to plagiarise, although she said such a policy would have to be open and based on clear criteria.

"If we do take up this tool, it's in our interests to tell authors upfront, not least because it's a warning not to bother sending in plagiarised work," she said.

Philippe Terheggen, director of journal development at Elsevier, said that across the group's 2,000 journals, 5,000 editors received 500,000 submissions every year.

Given the scale of the operation, he said, CrossCheck could be a valuable addition to the anti- plagiarism measures already used, but it would not replace peer review, he said.

"This would be an add-on, a help to the enormous efforts reviewers and editors make to deliver the level of integrity that they do," he said.

"If it is known that publishers are helping editors to detect plagiarism, authors may rethink their publishing ethics, and that is potentially more valuable than policing."

john.gill@tsleducation.com.