When Document Metadata Goes Awry: The Importance of Looking at the Bigger Picture

Most of us rely on data for telling an all-encompassing story; it fills in the gaps that we humans unknowingly create. We usually find data to be a credible source and something that can’t be misinterpreted. It’s an integral element of society. From our phones telling us about our screentime to our fitness trackers observing our heart rate over time. And let’s not forget that some banks can now reveal our, sometimes unsettling, spending habits. We call this helpful data, an insight.

Metadata, for example, provides a useful summary of basic information about another piece of data. With metadata, we can, quite easily, identify the author, creation date, last modified date, and editing time of a basic document. In the present day, law enforcement has even been found to use metadata as part of its criminal investigations.

This then begs the question of whether metadata alone is substantial enough to solve a crime. Or from our standpoint: is it sufficient enough to solve an academic misconduct investigation? Emily Gover, a former Curriculum Developer at Easybib, unexpectedly found herself wondering the same thing...

That time Emily authored an economics paper she never actually wrote.

In an all-too interesting and almost-unbelievable story, Emily was contacted by a panicked economics student from Colorado. She recounts the Twitter exchange from 2016:

“I got this direct message from a student that said: ‘This is really weird but I was wondering if you’ve written any economics papers? I’m being told by my professor that I’ve plagiarized this paper for my 300 level economics class, and I swear I’ve written everything myself, and I’ve no idea what’s happened... and he’s threatening to fail me! And if I don’t pass this, I won’t graduate.’

“So at first, I told him I definitely haven’t written any economics stuff. Eventually, after a few exchanges, I said, ‘I work at this company that does citations. Did you use Easybib or any other of our websites?’”

As with most academic assignments, this student was required to correctly format his paper. He did so using an APA title generator, called Citation Machine, owned by Easybib.

Emily went on to tell us, “He reached out to his professor and asked, ‘What was in your software that had Emily’s name on it?” and the professor said that it was in the document metadata.”

As a curriculum developer, Emily helped to create the content that formed Easybib’s Citation Machine. Unbeknown to her, the documents she supplied to her team had her name in its metadata. The student downloaded his title page from Easybib and began writing his paper in the same document. It’s as simple and as innocent as that. His economics professor, on the other hand, saw a red flag and was obligated to pursue an investigation.

So, is metadata alone substantial in solving an academic misconduct case?

When time is of the essence—particularly as an educator—technology lightens the load; it makes an academic misconduct investigation a whole lot quicker and easier. But one thing is for certain: technology can only help in this process. Machines can’t solve investigations for us, and manual input is always required.

Emily points out: “For all we know, the professor could have seen this before with students who went to an essay mill and spent $50 to have someone write their essay for them, thinking they could get away with it.

“Metadata gives you evidence of the origin of that document. But this was an outlier. I think it can be very easy for humans to become reliant on things. ‘Well, this robot’s telling me that this is what’s happening and so it’s got to be true!’ When, in reality, there are a lot more nuances that tech can’t pick up.”

What happened to the student from Colorado?

You can rest easy tonight. Emily informed us, “I wrote a statement that the student sent on to the professor. There was a confirmation to the school saying: ‘My name is Emily Gover, this is what I do. Your student legally used this tool. It’s just a weird snafu of metadata and technology thinking that it has something right when, in reality, there are a lot more nuanced details that algorithms and bots can’t pick up.’

“The student was super appreciative and told me a week later that everything passed after being threatened to be put on academic probation. Everything turned out fine and he graduated.”

A big thanks to Emily for telling her story. You can read Emily’s original writeup about this ‘weird snafu’ on her personal blog.

As academic misconduct evolves from plagiarism into contract cheating, at Turnitin, we’re trying as hard as ever to evolve, too. But, until robots walk among us, humans must always make the final decision.

So, for now, we have Turnitin Originality, which offers data-backed, multi-faceted insights (helpful data!) into whether students are doing their own work. With the use of forensic linguistic analysis, Natural Language Processing, and metadata, Turnitin Originality's technology helps instructors pull a number of singular pieces of evidence into one space to help educators review them as a whole. 

Turnitin Originality gathers all of the necessary pieces of evidence to help educators make informed and impartial judgments with confidence.