Detecting Plagiarism in Student Assignments using Source Code Analysis

Abstract

The issue of accurately detecting semantically equivalent code remains a current problem for students, teachers, and researchers. The goal of this article was to develop an alternative method for checking software code for plagiarism using abstract syntax trees. To achieve this, an analysis of existing modern methods of automatic comparison of original and modified program code for plagiarism detection was conducted, focusing on abstract syntax trees, approaches based on machine learning, and token-based analysis. Systems such as MOSS, JPlag, Codequiry, and Plaggie were reviewed. As a result of this work, it can be noted that the proposed methods have sufficiently high indicators in the range of 79-85% for semantically equivalent code and 93-95% similarity in cases demonstrating hash violation problems, which shows their potential effectiveness in solving various tasks related to the detection of software code plagiarism. The practical value of the research lies in the possibility of using the proposed method for checking small student assignments for originality. Future research on this topic could include adding functionality for document collection, reducing the time required for document verification, expanding the number of programming languages, and improving system efficiency while operating large code fragments.