Over 31% of submissions to the National Natural Science Foundation of China English-language publications show signs of copying, self-plagiarism, or copyright infringement, according to Helen Zhang, journal director at the Zhejiang University Press in China.
Zhang used the text analysis service CrossCheck to detect articles submitted to the Journal of Zhejiang University – Science (JZUS) A, JZUS B, and JZUS C for violations of any of four distinct types of plagiarism that she believes constitute academic misconduct: duplicate publication, self or team plagiarism, direct copying of methods section, and uncited or excessive extracts.
In an article published earlier this year in Learned Publishing, Zhang and her colleagues analyzed manuscripts submitted to the three journals between October 2008 and May 2009. They discovered evidence of plagiarism in 151 out of 622 papers, or about 22.8% of submissions during that seven-month period. In a recent opinion piece published in Nature, Zhang reports that continued data collection analysis shows that the problem is only escalating as the percentage of plagiarized submissions has increased to 31%.
Zhang says that in the face of mounting pressure to increase the length of their publication listing, there is an increasing trend for individual members of a research team to submit several short, insubstantial, and similar papers instead of one strong paper. But convincing research teams that this patchwork process is plagiarism can be a difficult task for editors. “At the beginning, they usually argue with us,” said Zhang, who believes many cases of self or team plagiarism is a result of the academic world’s emphasis on publishing quantity over quality. “We often ask them, ‘Why can’t your group organize a long and strong paper on this topic?’”
Offered by the nonprofit association CrossRef, the CrossCheck service is designed to help publishers identify plagiarism by comparing the content of a submitted paper to a continuously updated database of previously published work. The service received the 2008 Association of Learned and Professional Society Publishers Award for Publishing Innovation. But while CrossCheck can find a significant number of plagiarism incidents, it is not foolproof. For example, it cannot analyze figures and tables, so journal editors must rely on sources like Google or PubMed Central to round out their vigilance.
Zhang hopes the results of her analysis will encourage the academic world to reevaluate its priorities in order to reward original, thorough research and discourage a publishing frenzy that diminishes the resources of the journal and its readers. While the multi-step CrossCheck protocol lengthens the already lengthy submission process, she stresses that the additional time used to detect plagiarism is worth the benefits reaped from responsible academic journalism.
“It is a long-term mission to protect original scientific fruits and become more respectful of authors’ copyright,” said Zhang. “The mission of subjectively cultivating the human character, promoting cultured academic behavior, and objectively stopping academic misconduct cannot be completed in one day.”
1. Zhang, Helen (Yuehong). 2010. CrossCheck: an effective tool for detecting plagiarism. Learned Publishing 23: 9-14.
2. Zhang, Yuehong. 2010. Chinese journal finds 31% of submissions plagiarized. Nature 467:71, 271.