Last Friday, the Chronicle of Higher Education ran a story about the emerging anti-plagiarism software marketplace, with CrossRef’s CrossCheck spotlighted. It’s a good story that broadens nicely upon the CrossCheck angle.
Defining plagiarism is potentially fraught with difficulty, but it has been done:
Plagiarism is the deliberate attempt to deceive the reader through the appropriation and representation as one’s own the work and words of others. Academic plagiarism occurs when a writer repeatedly uses more than four words from a printed source without the use of quotation marks and a precise reference to the original source in a work presented as the author’s own research and scholarship. Continuous paraphrasing without serious interaction with another person’s views, by way or argument or the addition of new material land insights, is a form of plagiarism in academic work.
Four words? Pretty harsh! I think you could do that accidentally. How does CrossCheck deal with that?
When the system detects a degree of textual overlap between the submitted manuscript and items in the database, the system produces an originality report, which can be used by editorial staff in order to determine whether the overlap is benign, or a case of plagiarism.
OK, that’s better. Still, it will be interesting to see how well this works.
It also puzzles me a little as to how big a problem this is in STM communication. I believe it is a problem more in the humanities, and even by form (books and essays). Do STM publishers really need a major infrastructure deployment to deal with it?
I can see the problem it poses in university classes, where the opportunities for local plagiarism are rife, and the chance of being caught by a professor are slight. The distribution is just too small and local. But for STM publishers, is this using a hammer to kill a fly?
If plagiarism is eliminated from the system of scholarly research by software like CrossCheck and others, we will be left with an even more vexing and debilitating issue — fabrication.
And how do you create an automated BS detector?