PasteTrace: A Single Source Plagiarism Detection Tool For Introductory Programming Courses

Jesse McDonald, Scott Robertson, Anthony Peruma · 2025

Read PasteTrace: A Single Source Plagiarism Detection Tool For Introductory Programming Courses by Jesse McDonald, Scott Robertson, Anthony Peruma (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-17355.txt.

Introduction

McDonald, Robertson, and Peruma built a plagiarism detection tool in twenty twenty-five aimed at first programming courses. New coders who struggle may copy work, and generic detectors miss how copying actually happens in a classroom.

Their tool, PasteTrace, lives inside the coding editor students already use. It watches the writing process itself, recording events such as large pastes that suggest outside code arriving all at once. The authors trialed it in two real introductory courses. It surfaced telling patterns in student behavior and caught forms of plagiarism that a well-known comparison tool missed. Because it reads the process, not just the final file, it can flag suspicious work even when the code itself has been disguised. The tool is open source, and the team shares a demo video and the trial data.

Catalog record

CREATORJesse McDonald, Scott Robertson, Anthony Peruma
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-17355

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

PasteTrace: A Single Source Plagiarism Detection Tool For Introductory Programming Courses Jesse McDonald, Scott Robertson, Anthony Peruma Abstract Introductory Computer Science classes are important for laying the foundation for advanced programming courses. However, students without prior programming experience may find these courses challenging, leading to difficulties in understanding concepts and engaging in academic dishonesty such as plagiarism. While there exists plagiarism detection techniques and tools, not all of them are suitable for academic settings, especially in introductory programming courses. This paper introduces PasteTrace, a novel open-source plagiarism detection tool designed specifically for introductory programming courses. Unlike traditional methods, PasteTrace operates within an Integrated Development Environment that tracks the student’s coding activities in real-time for evidence of plagiarism. Our evaluation of PasteTrace in two introductory programming courses demonstrates the tool’s ability to provide insights into student behavior and detect various forms of plagiarism, outperforming an existing well-established tool. A video demonstration of PasteTrace and its source code, and case study data are made available at https://doi.org/10.6084/m9.figshare.27115852 I Introduction Introductory Computer Science classes serve as the initial exposure to programming for many students, playing a vital role in building the foundation for subsequent higher-level courses [ 1 ] . It is through these courses students learn about the fundamentals of programming, such as variables, data types, control structures, and functions, which they build upon in their subsequent higher-level courses [ 2 , 3 ] . Moreover, the importance of these courses extends beyond students learning about programming language semantics and syntax. Through these courses, students also develop and practice problem-solving, creativity, and logical reasoning skills, which they can apply to their programming assignments [ 4 ] . However, these courses can be challenging for many students, especially those with no prior programming experience. These challenges range from syntactic difficulties to conceptual misunderstandings as well as deficits in problem-solving, code organization, and debugging [ 5 ] . Consequently, students may unintentionally or intentionally engage in academic dishonesty, such as plagiarism [ 6 ] . Plagiarism is particularly troublesome in introductory classes. Since an introductory class is often a student’s first experience of programming, if a student passes an introductory class by plagiarizing, then it is likely that the student does not understand part or all of the course. If students are not corrected for plagiarizing in their introductory programming classes, they may develop a false sense of their abilities. This lack of understanding due to plagiarism could leave them unprepared for more advanced courses and future career challenges. I-A Goal In the field of computing education, several techniques and tools have been developed to identify different types of plagiarism [ 7 , 8 ] . While these tools are valuable, not all of them are suitable for academic settings, especially in the context of introductory programming courses. Some of these tools are proprietary and closed-source, making it challenging for institutions to assess their accuracy and reliability. Additionally, certain plagiarism detection tools require the transfer of files to external services, raising concerns about potential violations of student privacy laws. The goal of our research is to provide an effective and reliable open-source code plagiarism detection tool specifically tailored for academic environments, particularly in introductory programming courses. To this extent, we present PasteTrace, a single source plagiarism detection tool For introductory programming courses . II Background Plagiarism is generally defined as “The practice of taking someone else’s work or ideas and passing them off as one’s own.” [ 9 ] However, for this paper, we define plagiarism as “ copying someone else’s code to complete an assignment .” We exclude copying ideas for several reasons, mainly because it is difficult to determine if an idea has been copied or just independently rediscovered; the general culture of computer science is to iterate on others ideas (often without attribution [ 10 ] ); and because at the introductory level, all the required ideas have been thought of by so many people that determining an origin would be next to impossible. Additionally, we exclude the possibility a student could “cite” their copied code to avoid plagiarism because the introductory level involves teaching coding, not copying existing code. II-A Existing Techniques for Detecting Code Plagiarism The general method of plagiarism can be summarized as follows: Given a set A of submissions, find a set P such that \forall p\in A, if \exists q\in A such that q\not=p and D(p,q) , then p\in P and q\in P for some plagiarism detection function D . However, defining D effectively can be challenging [ 11 ] . Researchers at Stanford University developed a program called MOSS (Measure of Software Similarity), which is capable of giving each assignment pair a similarity score [ 12 ] . This approach is robust against name changes, code reorganization, and white space changes by analyzing a semi-random subset of k-gram hashes based on each assignment where k has been manually tuned for each language. Others have attempted to define D by ignoring syntax and exclusively using code semantics [ 13 ] so that two pieces of code match when they functionally do the same thing. There are also many other efforts to define D using a machine learning model [ 14 , 15 , 16 , 17 ] . Additionally, there are also detection models that attempt to exclusively detect single LLMs output [ 18 ] . However, these are not particularly reliable at present [ 19 ] and have a high false positive rate [ 20 ] . AlSallal et al. [ 21 ] used machine learning to detect stylistic variation in essays. However, in our scenario, this approach would not be ideal for introductory-level students who may not yet have a defined “style.” III PasteTrace III-A Single Source Plagiarism Detection PasteTrace is an Integrated Development Environment (IDE) based on the Processing IDE. Processing is an open-source IDE for Java that is designed to be easy to learn even for non-programmers [ 22 ] ; as such, it is ideal for our application. When the modified IDE first launches, a persistent UUID file is created on the machine that installed it. This UUID acts as a machine or user ID, is referred to as the InstallID , and is assumed to be constant throughout the class. Additionally, whenever a project is created with the IDE, a second UUID is created that is unique to that project. We refer to this as the ProjectID . Both UUIDs are saved in a special hidden metadata comment in the program source file. We will refer to this as the metaComment . Whenever a file is opened, the IDE checks the InstallID of the file and machine. If they differ, the IDE notes the new InstallID in the metaComment in an ordered list called the InfectionStack . A similar stack is also used to track paste events. As these stacks are effectively a single stack with a way to discriminate between event types, we will use InfectionStack to refer to both. Additionally, when any part of the code is copied, zero-width spaces (U+200B) are interspersed with the normal letters to encode as much data as possible. The exact amount of data varies based on the size of the copy, but

Read the complete text