Source Code Plagiarism Detection Research Articles

Detecting Source Code Plagiarism in Student Assignment Submissions Using Clustering Techniques

In pragmatic courses, graduate students are required to submit programming assignments, which have been susceptible to various forms of plagiarism. Detecting counterfeited code in an academic setting is of paramount importance, given the prevalence of publications and papers. Plagiarism, defined as the unauthorized replication of written work without proper acknowledgment, has become a critical concern with the advent of information and communication technology (ICT) and the widespread availability of scholarly publications online. However, the extensive use of freeware text editors has posed challenges in detecting source code plagiarism. Numerous studies have investigated algorithms for revealing different types of plagiarism and detecting source code plagiarism. In this research, we propose an innovative strategy that combines TF-IDF (Term Frequency-Inverse Document Frequency) modifications with K-means clustering, achieving a remarkable precision rate of 99.2%. Additionally, we explore the hierarchical clustering method, which estimates an even higher precision rate of 99.5% compared to previous techniques. To implement our approach, we utilize the Python programming language along with relevant libraries, providing a robust and efficient system for source code plagiarism detection in student assignment submissions.

Classification feature sets for source code plagiarism detection in Java

In programming learning environments, the pressure of delivering many programming assignments makes plagiarism the easiest solution. This highly threatens the learning process; therefore, the need of an automatic, fast, and accurate detection of source code plagiarism becomes essential. To detect whether a pair of Java files is plagiarized, this paper proposes four classification feature sets: (i) structural histogram features, histogram-based features for summarizing similarity matrices; (ii) lexical per-class features, extracted from a lexical similarity matrix between the classes of the two compared files based on character 3-grams; (iii) structural counting features, twelve counting features representing the code structure; and (iv) modified original features: a set of modifications on the features of the used baseline. The results show that the best feature sets in F-measure are the structural histogram features and the lexical per-class features combined, which improve the F-measure by 4% compared to the baseline. The added features slow down the execution time. However, it is still efficient, given that it can classify 70k pairs in 23 min. In addition, we partially re-annotated the SOurce COde Re-use dataset. After the re-annotation, the F-measure of both the baseline and our work is improved, and our work achieves an F-measure of 93.6%, which is 7.5% higher than the new F-measure of the baseline. In addition, some remarks and recommendations are provided for using the SOurce COde Re-use dataset as a benchmark.

Comparison of parallel central processing unit‐ and graphics processing unit‐based implementations of greedy string tiling algorithm for source code plagiarism detection

SummaryMassive‐enrollment computing courses often involve some practical training through programming assignments and projects that are frequent targets for plagiarism. Source code similarity detection tools are used to prevent such misbehavior. Parallel processing has recently become a viable technique for speeding up the processing of large workloads. This article examines the parallelization of a source code similarity detection method based on the greedy string tiling and Karp–Rabin algorithms. Both CPU and GPU parallelization approaches are discussed. The CPU implementation uses Pthreads, whereas the GPU implementation employs CUDA. Depending on the evaluated dataset which consists of real student assignment codes, speedups of up to seven times over the sequential version of the code are achieved. Evaluation results on both platforms are compared and discussed in detail.

Identifying plagiarised programming assignments based on source code similarity scores

ABSTRACT Background and Context Source code plagiarism is a common occurrence in undergraduate computer science education. Many source code plagiarism detection tools have been proposed to address this problem. However, such tools do not identify plagiarism, nor suggest what assignment submissions are suspicious of plagiarism. Source code plagiarism detection tools simply evaluate and report the similarity of assignment submissions. Detecting plagiarism always requires additional human intervention. Objective This work presents an approach that enables the automated identification of suspicious assignment submissions by analysing similarity scores as reported by source code plagiarism detection tools. Method Density-based clustering is applied to a set of reported similarity scores. Clusters of scores are used to incrementally build an association graph. The process stops when there is an oversized component found in the association graph, representing a larger than expected number of students plagiarising. Thus, the constructed association graph represents groups of colluding students. Findings The approach was evaluated on data sets of real and simulated cases of plagiarism. Results indicate that the presented approach can accurately identify groups of suspicious assignment submissions, with a low error rate. Implications The approach has the potential to aid instructors in the identification of source code plagiarism, thus reducing the workload of manual reviewing.

Identifying Plagiarised Programming Assignments with Detection Tool Consensus

Source code plagiarism is a common occurrence in undergraduate computer science education. Many source code plagiarism detection tools have been proposed to address this problem. However, most of these tools only measure the similarity between assignment submissions, and do not actually identify which are suspicious of plagiarism. This work presents a semi-automatic approach that enables the indication of suspicious assignment submissions by analysing source code similarity scores among the submissions. The proposed approach seeks the consensus of multiple source code plagiarism detection tools in order to identify program pairs that are consistently evaluated with high similarity. A case study is presented to demonstrate the use of the proposed approach. The results of this case study indicate that it can accurately identify assignment submissions that are suspicious of plagiarism.

Dolos: Language‐agnostic plagiarism detection in source code

AbstractBackgroundLearning to code is increasingly embedded in secondary and higher education curricula, where solving programming exercises plays an important role in the learning process and in formative and summative assessment. Unfortunately, students admit that copying code from each other is a common practice and teachers indicate they rarely use plagiarism detection tools.ObjectivesWe want to lower the barrier for teachers to detect plagiarism by introducing a new source code plagiarism detection tool (Dolos) that is powered by state‐of‐the art similarity detection algorithms, offers interactive visualizations, and uses generic parser models to support a broad range of programming languages.MethodsDolos is compared with state‐of‐the‐art plagiarism detection tools in a benchmark based on a standardized dataset. We describe our experience with integrating Dolos in a programming course with a strong focus on online learning and the impact of transitioning to remote assessment during the COVID‐19 pandemic.Results and ConclusionsDolos outperforms other plagiarism detection tools in detecting potential cases of plagiarism and is a valuable tool for preventing and detecting plagiarism in online learning environments. It is available under the permissive MIT open‐source license at https://dolos.ugent.be.ImplicationsDolos lowers barriers for teachers to discover, prove and prevent plagiarism in programming courses. This helps to enable a shift towards open and online learning and assessment environments, and opens up interesting avenues for more effective learning and better assessment.

Enhanced Semantic Similarity Detection of Program Code Using Siamese Neural Network

Even though there are various source code plagiarism detection approaches, most of them are only concerned with lexical similarities attack with an assumption that plagiarism is only conducted by students who are not proficient in programming. However, plagiarism is often conducted not only due to student incapability but also because of bad time management. Thus, semantic similarity attacks should be detected and evaluated. This research proposes a source code semantic similarity detection approach that can detect most source code similarities by representing the source code into an Abstract Syntax Tree (AST) and evaluating similarity using a Siamese neural network. Since AST is a language-dependent feature, the SOCO dataset is selected which consists of C++ program codes. Based on the evaluation, it can be concluded that our approach is more effective than most of the existing systems for detecting source code plagiarism. The proposed strategy was implemented and an experimental study based on the AI-SOCO dataset revealed that the proposed similarity measure achieved better performance for the recommendation system in terms of precision, recall, and f1 score by 15%, 10%, and 22% respectively in the 100,000 datasets. In the future, it is suggested that the system can be improved by detecting inter-language source code similarity.

Evaluating the robustness of source code plagiarism detection tools to pervasive plagiarism-hiding modifications

Source code plagiarism is a common occurrence in undergraduate computer science education. In order to identify such cases, many source code plagiarism detection tools have been proposed. A source code plagiarism detection tool evaluates pairs of assignment submissions to detect indications of plagiarism. However, a plagiarising student will commonly apply plagiarism-hiding modifications to source code in an attempt to evade detection. Subsequently, prior work has implied that currently available source code plagiarism detection tools are not robust to the application of pervasive plagiarism-hiding modifications. In this article, 11 source code plagiarism detection tools are evaluated for robustness against plagiarism-hiding modifications. The tools are evaluated with data sets of simulated undergraduate plagiarism, constructed with source code modifications representative of undergraduate students. The results of the performed evaluations indicate that currently available source code plagiarism detection tools are not robust against modifications which apply fine-grained transformations to the source code structure. Of the evaluated tools, JPlag and Plaggie demonstrates the greatest robustness to different types of plagiarism-hiding modifications. However, the results also indicate that graph-based tools, specifically those that compare programs as program dependence graphs, show potentially greater robustness to pervasive plagiarism-hiding modifications.

Academic Source Code Plagiarism Detection by Measuring Program Behavioral Similarity

Source code plagiarism is a long-standing issue in tertiary computer science education. Many source code plagiarism detection tools have been proposed to aid in the detection of source code plagiarism. However, existing detection tools are not robust to pervasive plagiarism-hiding transformations, and as a result can be inaccurate in the detection of plagiarised source code. This article presents BPlag, a behavioural approach to source code plagiarism detection. BPlag is designed to be both robust to pervasive plagiarism-hiding transformations, and accurate in the detection of plagiarised source code. Greater robustness and accuracy is afforded by analysing the behaviour of a program, as behaviour is perceived to be the least susceptible aspect of a program impacted upon by plagiarism-hiding transformations. BPlag applies symbolic execution to analyse execution behaviour and represent a program in a novel graph-based format. Plagiarism is then detected by comparing these graphs and evaluating similarity scores. BPlag is evaluated for robustness, accuracy and efficiency against 5 commonly used source code plagiarism detection tools. It is then shown that BPlag is more robust to plagiarism-hiding transformations and more accurate in the detection of plagiarised source code, but is less efficient than compared tools.

Automated System for Source Code Plagiarism Detection

Plagiarism of code is a serious issue in today’s era. Plagiarism refers to the use of someone’s data, language and writing without proper acknowledgement of the original source. Our project basically aims to detect plagiarism in subsequent code submissions made by students. We are hosting a front-end website from where the user will be uploading the source code, after this in the website the code submissions will be made to check the extent of plagiarism in the code submissions made by students. We will check the extent of plagiarism by checking multiple parameters like number of lines of code, number of functions used, number of loops used and the extent to which keywords match in subsequent codes.

Particle swarm optimisation-based source code plagiarism detection approach using non-negative matrix factorisation algorithm

Source code plagiarism is easy to do the task, but very difficult to detect without proper tool support. Various source code similarity detection systems have been developed to help detect source code plagiarism. Numerous efforts have been made in the literature to introduce an efficient source code detection approach with less time complexity and accurate classification of plagiarised codes. However, there exists a tradeoff amongst the less complexity and high accuracy. In a similar way, this paper likewise attempted to build a framework to detect the plagiarised codes from the source code corpus. This approach employed an intelligent swarm optimisation algorithm known as PSO in the detection phase and robust matrix factorisation algorithm known as non-negative matrix factorisation based on alternative least square (ALS) algorithm for reduction of features from the sparse matrix. Depending on the implementation, ALS is very fast and significantly less work than an SVD implementation. The experimental results showed that it has good performance compared to the other existing approaches such as precision and recall.

Particle swarm optimisation-based source code plagiarism detection approach using non-negative matrix factorisation algorithm

Source code plagiarism is easy to do the task, but very difficult to detect without proper tool support. Various source code similarity detection systems have been developed to help detect source code plagiarism. Numerous efforts have been made in the literature to introduce an efficient source code detection approach with less time complexity and accurate classification of plagiarised codes. However, there exists a tradeoff amongst the less complexity and high accuracy. In a similar way, this paper likewise attempted to build a framework to detect the plagiarised codes from the source code corpus. This approach employed an intelligent swarm optimisation algorithm known as PSO in the detection phase and robust matrix factorisation algorithm known as non-negative matrix factorisation based on alternative least square (ALS) algorithm for reduction of features from the sparse matrix. Depending on the implementation, ALS is very fast and significantly less work than an SVD implementation. The experimental results showed that it has good performance compared to the other existing approaches such as precision and recall.

A Language-Independent Library for Observing Source Code Plagiarism

Background: Most source code plagiarism detection tools are not modifiable. Consequently, when a modification is required to be applied, a new detection tool should be created along with it. This could be a problem as creating the tool from scratch is time-inefficient while most of the features are similar across source code plagiarism detection tools.Objective: To alleviate researchers' effort, this paper proposes a library for observing two plagiarism-suspected codes (a feature which is similar across most source code plagiarism detection tools).Methods: Unique to this library, it is not constrained by the selected programming language for development. It is executed from command line, which is supported by most programming languages.Results: According to our evaluation, the library is integrable and functional. Moreover, the library can enhance teaching assistants' accuracy and reduce the tasks' completion time.Conclusion: The library can be beneficial for the development of source code plagiarism detection tools since it is integrable, functional, and helpful for teaching assistants.Keywords:Language independency, Plagiarism detection, Reusable library, Source code, Tool development

Source Code Plagiarism Detection in Academia with Information Retrieval: Dataset and the Observation

Source code plagiarism is an emerging issue in computer science education. As a result, a number of techniques have been proposed to handle this issue. However, comparing these techniques may be challenging, since they are evaluated with their own private dataset(s). This paper contributes in providing a public dataset for comparing these techniques. Specifically, the dataset is designed for evaluation with an Information Retrieval (IR) perspective. The dataset consists of 467 source code files, covering seven introductory programming assessment tasks. Unique to this dataset, both intention to plagiarise and advanced plagiarism attacks are considered in its construction. The dataset's characteristics were observed by comparing three IR-based detection techniques, and it is clear that most IR-based techniques are less effective than a baseline technique which relies on Running-Karp-Rabin Greedy-String-Tiling, even though some of them are far more time-efficient.

Using Stack Overflow content to assist in code review

SummaryAn essential goal for programmers is to minimize the cost of identifying and correcting defects in source code. Code review is commonly used for identifying programming defects. However, manual code review has some shortcomings: (1) it is time‐consuming and (2) outcomes are subjective and depend on the skills of reviewers. An automated approach for assisting in code reviews is thus highly desirable. We present a tool for assisting in code review and results from our experiments evaluating the tool in different scenarios. The tool leveraged content available from professional programmer support forums (eg, StackOverflow.com) to determine potential defectiveness of a given piece of source code. The defectiveness is expressed on the scale of {Likely defective, neutral, unlikely to be defective}. The basic idea employed in the tool is (1) to identify a set P of discussion posts on Stack Overflow such that each p∈P contains source code fragment(s), which sufficiently resemble the input code C being reviewed, and (2) to determine the likelihood of C being defective by considering all p∈P. A novel aspect of our approach is to use document fingerprinting for comparing two pieces of source code. Our choice of document fingerprinting technique is inspired by source code plagiarism detection tools where it has proven to be very successful. In the experiments that we performed to verify the effectiveness of our approach, source code samples from more than 300 GitHub open‐source repositories were taken as input. An F1 score of 0.94 has been achieved in identifying correct/relevant results.

Source-code Similarity Detection and Detection Tools Used in Academia

Teachers deal with plagiarism on a regular basis, so they try to prevent and detect plagiarism, a task that is complicated by the large size of some classes. Students who cheat often try to hide their plagiarism (obfuscate), and many different similarity detection engines (often called plagiarism detection tools) have been built to help teachers. This article focuses only on plagiarism detection and presents a detailed systematic review of the field of source-code plagiarism detection in academia. This review gives an overview of definitions of plagiarism, plagiarism detection tools, comparison metrics, obfuscation methods, datasets used for comparison, and algorithm types. Perspectives on the meaning of source-code plagiarism detection in academia are presented, together with categorisations of the available detection tools and analyses of their effectiveness. While writing the review, some interesting insights have been found about metrics and datasets for quantitative tool comparison and categorisation of detection algorithms. Also, existing obfuscation methods classifications have been expanded together with a new definition of “source-code plagiarism detection in academia.”

Source code plagiarism detection with low-level structural representation and information retrieval

Low-level approach is a recent way for source code plagiarism detection. Instead of relying on source code tokens, it relies on low-level structural representation resulted from compiling given source codes. However, to date, existing low-level approaches are unsuitable for handling large-sized source codes; their comparison takes either quadratic or cubic time complexity. This paper presents a low-level approach with a more-efficient comparison; the comparison only takes linear time complexity with the help of Cosine Correlation in Vector Space Model. According to our evaluation, three findings can be deducted. First, proposed approach can be more efficient and effective than the state-of-the-art approach. Second, low-level structural representation can be more effective and efficient than source code token sequence as a medium for source code plagiarism detection. Third, subsequence length (which is related to n in n-gram) can be inversely proportional to effectiveness without significant impact on efficiency.

Automated Hints Generation for Investigating Source Code Plagiarism and Identifying The Culprits on In-Class Individual Programming Assessment

Most source code plagiarism detection tools only rely on source code similarity to indicate plagiarism. This can be an issue since not all source code pairs with high similarity are plagiarism. Moreover, the culprits (i.e., the ones who plagiarise) cannot be differentiated from the victims even though they need to be educated further on different ways. This paper proposes a mechanism to generate hints for investigating source code plagiarism and identifying the culprits on in-class individual programming assessment. The hints are collected from the culprits’ copying behaviour during the assessment. According to our evaluation, the hints from source code creation process and seating position are 76.88% and at least 80.87% accurate for indicating plagiarism. Further, the hints from source code creation process can be helpful for indicating the culprits as the culprits’ codes have at least one of our predefined conditions for the copying behaviour.

A CSA based Source Code Plagiarism Detection Approach using Sparse Principle Component Analysis

A CSA based Source Code Plagiarism Detection Approach using Sparse Principle Component Analysis

ES‐Plag: Efficient and sensitive source code plagiarism detection tool for academic environment

AbstractSource code plagiarism detection using Running‐Karp‐Rabin Greedy‐String‐Tiling (RKRGST) is a common practice in academic environment. However, such approach is time‐inefficient (due to RKRGST's cubic time complexity) and insensitive (toward token subsequence rearrangement). This paper proposes ES‐Plag, a plagiarism detection tool featured with cosine‐based filtering and penalty mechanism to handle aforementioned issues. Cosine‐based filtering mitigates time‐inefficiency by excluding non‐potential pairs from RKRGST comparison; while penalty mechanism mitigates insensitivity by reducing the number of matched tokens with the number of matched subsequences prior similarity normalization. In addition to issue‐solving features, ES‐Plag is also featured with project‐based input, colorized adjacency similarity matrix, matched token highlighting, and various similarity algorithms (e.g., Cosine Similarity and Local Alignment). Three findings can be deducted from our evaluation. First, cosine‐based filtering boosts up time efficiency with a trade‐off in effectiveness. Second, penalty mechanism enhances sensitivity even though its improvement in terms of effectiveness is quite limited. Third, ES‐Plag's features are beneficial for examiners.