Plagiarism detection 2.0: Towards detection of contract cheating and unauthorized assistance in school coursework

The other day me, and my friend (who is an instructor at a local university) were having a discussion on student collusion. He was telling me how he could “feel” that some of his students were just copying off from other’s work. But he just couldn’t prove it.

To add to the problem, since he teaches electronics engineering; the student’s have to submit logic diagrams. Since there is often only one correct answer, even if he knows that they copied from each other, he can’t really accuse them for having the right answer. Unless, they make the same mistakes. Which he said was improbable to happen.

Being interested in software development, I thought of writing “an app for that”. We decided to try a little experiment. I asked him, “Imagine if you gave them a document, and that document knew where it was, who accessed it, when it was accessed and what text was taken out of it and put into it. Theoretically, wouldn’t that document know who colluded? Since it was there and it saw the collusion happen. Apart from the student himself, the document is the only witness to the ‘crime’?”. Needless to say, we laughed for a few minutes. And, just for laughs he said he’d like to see that happen.

So, I built him a little desktop application, and a special filetype to distribute to his class. The interesting part was that the filetype worked with Microsoft Word, so there was no learning curve for the students. The software worked in the background, and recorded the “digital” footprints I mentioned earlier. A key thing here was that we asked his students to work ONLY on the particular files we provided (every student was given a separate one).

After the submission date, me and him sat together to look at the results. Before we looked at the reports, I asked him “Who do you think colludes?” and he named me three students. I said okay. So we checked their documents first. And, was he ever so right.

The results

Their documents showed that 100% of its contents were copied from some other file. Moreover, the student had only spent 45 seconds working on their paper. Which was enough evidence that they didn’t really work on it. All they did was open their document and copied from a document and closed it. Examining the contents of the file revealed that the text in the paper was actually copied from the paper of another student (as shown in the screenshot below).

What was most amazing to see was that even the diagram the student copied into his document was highlighted to be copied from the work of his peer.

Needless to say, my friend was very happy that he now had concrete evidence to have a talk with the student. And we generalized that this approach could be used to help screen instances of contract cheating as well.

How it works

A key part is limiting the file a student submits their work on (before submission). If a student is restricted to work on a single file and that file is trained to observe and record suspicious activity, screening for instance of contract cheating/unauthorized assistance becomes much easier. The question then becomes, what constitutes as “suspicious” activity?

If you look at the report I linked to above, you will find these data points: the time taken to complete the paper, the locations the document was accessed from, the copy/paste activity performed on the document and a unique PC ID. Once we have all this data, it becomes easier to ascertain whether a student contract cheated by looking at what locations the document was accessed from and what changes were added to it and from where. The time related metric is also helpful since it shows how much effort the student put into his work.

Call for Action

I assume this is a problem a lot of educators are facing today. And it is pretty evident that existing tools are no match for this problem. It would make me really happy if the system could do for others, what it did for my friend.

I’ve decided to make it available for free to educators who want to learn if someone in their class is using contract cheating services. You can express your interest by filling out a form here or by e-mailing me at wasified@gmail.com. Let’s work together to help uncover uncover contract cheating, and tip technology’s scale back to equilibrium.