Detecting Plagiarism based on the Creation Process

To this date, all methodologies for detecting plagiarism have focused on investigating the final digital “outcome”, eg. a document or source code. Our novel approach takes the creation process into account using logged events. We look at interaction logs of an author with the software used for creation of the work. Detection relies on comparing histograms of command usages of multiple logs. A work is classified as plagiarism, if its log deviates too much from logs of “honestly created” works or if its log is too similar to another log. The technique supports detecting plagiarism for digital outcomes stemming from unique tasks such as thesis as well as equal tasks such as assignments where the same problem sets are solved by many students. Evaluation focuses on the latter case using collected logs by an interactive development environment (IDE) from more than 60 students for three programming assignments.