Department of Computer Science & Software Engineering
The University of Western Australia
Last modified: 17 July 2023
Modified By: Michael Wise
Tool 1: document analysis tool When a student submits an assignment in the form of a Microsoft Word Document that was actually written by someone else various aspects of the document's metadata potentially provide clues that may help substantiate an allegation of cheating. It is often the case that when students obtain an essay or similar written by another person they lightly edit the document, for example, adding their name and institution. At times, it is possible that document Properties are wiped or provide insufficient data to conclude how the work was written. However, forensic techniques exist that can dig deeper into documents when they are unzipped as XML files. See: this https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7324152.
There are various clues in that can be drawn from the XML file that indicate potential cheating as outlines here https://doi.org/10.1007/s10805-019-09358-w and here https://doi.org/10.1007/978-3-031-12680-2_12
What I am proposing as a tool for the students to develop is a program (site etc) into which an MS Word document can be dropped or uploaded. The tool would then do the work of converting the file to a zip format, unzipping the file to its XML constituent parts, and analysing the relevant constituent parts to provide evidence to assess in cases of suspected cheating such as counts of revision identifiers, ratio of edited to words, versions of Word used, languages used in the document
In short, the tool would unzip and automate the analysis of cheating-relevant information in word documents. A really sophisticated version would also conduct a web search for any identified author or website author to see whether these are known cheating providers.
Tool 2 This idea is simpler than the first. When students take online tests in LMS (Blackboard)it is possible to download a log the test activity that includes IP addresses from where the test was accessed, date and time, etc. This log is downloadable as a comma delimited file (*.csv), that people typically open in Excel.
When students outsource test completion to third parties well-known large-scale cheating providers are often off-shore in Kenya, China, Ukraine, and India or Pakistan. The second tool I propose would automatically search the IP column in the online test log and flag (highlight) suspicious activities such as logins from Kenya, logins from multiple IP addresses for the same student on repeatable tests, logins from discrepant locations at times that would make travel between the locations impossible, logins at unusual times (e.g. 12am-5am). Incorporating IP lookups into this process should be able to show whether addresses are VPNs or not.
An excellent tool would rank suspicious activity by counting and indicating when multiple red flags are apparent on the same test-taking incident (e.g., IP login from in Perth at 12am, followed by IP in Kenya, test taken at 3am).
| Department of Computer Science & Software Engineering The University of Western Australia Last modified: 17 July 2023 Modified By: Michael Wise |
|