Duplication and plagiarism have dogged publishers and researchers for centuries. In the predigitized world, offenders were caught only by astute readers in the same field. Today, technology simultaneously complicates and simplifies the issue.
Technology, such as word-processing programs, makes it ever so easy to cut and paste content and move it from one place to another. And digitization of scholarly content makes it easy to find content online that may be a little too similar.
Today, there are tools for researchers and publishers to use in weeding out the overlap before time and money are invested in publication.
CrossRef, with its vast database of articles, saw the need to establish a way for publishers to check submitted manuscripts for overlap. In 2008, working with the inventors of the popular academic site TurnItIn, the people at CrossRef developed CrossCheck, a tool for detecting similarities in manuscripts. Publishers must sign an agreement with CrossCheck to allow the software company iParadigms to “crawl” full text; this constitutes a CrossCheck deposit. Publishers that use the software, called iThenticate, upload their full-text manuscripts, and a search of all CrossCheck deposits is conducted. A similarity report is provided within a few minutes.
What Publishers Are Finding
Similarity screening opens all kinds of issues for publishers. At what stage do you screen papers? Who will do the screening? Who will review the similarity report? How easy is it to read? Can you really establish a threshold for similarity? How much time will it take? What will we do when we find substantial overlap? A recent CSE webinar looked at two publishers that screen papers through CrossCheck.
Carissa Gilman, managing editor of Cancer and Cancer Cytopathology, reported that in 2011, about 1.8% of submissions were rejected because of issues noted in similarity reports. Like many publishers, the American Cancer Society (ACS) uses the reports as a teaching tool for authors.
Authors with minor instances of recycled text or self-plagiarism, particularly in the introduction and methods sections, are given an opportunity to rewrite their papers and resubmit. ACS shares the similarity reports with authors. Gilman noted that the most common responses from authors are anger and panic. Her advice was to make thoughtful decisions that can be easily defended, to have the similarity reports accessible in case the authors call and want to go over them, and to make authors understand that your goal is not to get them fired.
ACS screens all new submissions to avoid wasting editor and reviewer time. Gilman said that performing the checks takes 2–10 minutes of staff time for each submission.
Heidi Vermette, a production editor of Obstetrics & Gynecology, reported that the journal conducts the checks just before acceptance. She found that the process takes an average of about 20 minutes per manuscript.
Most commercial online-submission systems allow manuscripts to have checks done automatically. But just getting a report and a similarity number is not enough. Both Gilman and Vermette stressed the importance of manually reviewing each report and categorizing problems as redundant (self-plagiarism), minor copying of short phrases (sloppy paraphrasing and referencing), or clear plagiarism.
Although CrossCheck is a powerful tool for publishers, it has some limitations. iThenticate does not check figures or tables, only English-language content is checked, and changes in spelling or hyphenation will break up what would be large chunks of text overlap.
It must also be noted that the search is only as good as the database of deposited articles. If a specific field is not well covered, journals in the field may be missing many publications to check against.
More information about CrossCheck can be found at www.crossref.org/crosscheck.
ANGELA COCHRAN is director of journals at The American Society of Civil Engineers, Reston, VA.
