Angela Carbone (Monash) gave a presentation on a plagiarism detection tool that builds upon several years of work (student projects) in building a moodle plugin to detect code similarity. I have a specific interest in this project as the CSER group will be working the Ange over the next year or so in trialling this tool in our courses.
Ange’s master student had looked at what current academic practices are in relation to plagiarism, which mostly focus on preventative mechanisms, such as training and assessment practices that discourage students from having to plagiarise. I was a little surprised that they didn’t find any code similarity detection software in use, as we have been using MOSS from Stanford for a number of years. However, they indicated that the reasons that academics don’t use this tool or other related tools, is that they are hard to use by the academics, and also cannot be used by students to help understand plagiarism. I agree with both of these, but hadn’t realised it was such an obstacle for academics. Turning this into a moodle plugin is a great idea as it will build upon an existing, well established tool, but make it easier and more effective for both students and academics to use. In turn, this means we can more easily undertake more research as to whether these kinds of plagiarism tools are useful in helping guide students into appropriate behaviours.
The way that the tool works is to enable students to submit code through moodle assignment uploading mechanisms. You can set scanning dates where the code submissions are scanned for similarity. The options are in the lecturer’s hands, but you can choose to set up scanning prior to the final submission date so that students can see themselves whether there has been any code similarity detected.
The initial studies that Ange has done have identified that their may be an increase in understanding of academic integrity issues, but that it may also raise anxiety issues. Also, academics thought that the prior to submission scanning might help students develop better disguises for their code. I think this concern often comes up when you discuss this kind of system, but there is always the question that if a student understands the code enough to change it enough so that similarity is not detected then why would they have plagiarised in the first place?
I’m looking forward to seeing how this works for us!
There was an interesting question relating to whether the similarity detection works across the current course, or a database of submission such as how Turn It In works. MOSS, from my understanding, works only on the submissions that are given to it, so it would work for previous course comparison if they were also included in the submissions. One of the problems we have in plagiarism is in purchasing solutions – it doesn’t show up in detection systems like this, but sometimes it will turn it up in Turn It In, but we do have issues with students hiring others to develop solutions for them – but this is a different problem.
We also discussed the thorny issue of what is plagiarism in non text-based environments. We often want our students to use existing algorithms, what defines plagiarism here? Do we need to come up with citation mechanisms for code? And we also don’t want to stop students from sharing ideas!