Normal view MARC view ISBD view

Python code similarity detection using abstract syntax tree with k-means algorithm / Venn P. Delos Santos, John Christian M. Gava, and Marc Christian D. Tumaneng

By:

Delos Santos, Venn P [author.]

Contributor(s):

Material type: Text

TextDescription: 166 leavesContent type:

text

Media type:

unmediated

Carrier type:

volume

Dissertation note: Undergraduate Thesis (BSCS) - Computer Science. Camarines Sur Polytechnic Colleges, 2024. Abstract: This study presents a hybrid methodology for detecting source code similarity in Python programming language. Integrating Abstract Syntax Tree (AST) analysis for structural similarity evaluation, the SimHash algorithm for textual similarity assessment, and the K-Means clustering algorithm for grouping Python codes according to their similarity fea-tures. The evaluation of the source code similarity detection was thoroughly conducted us-ing a dataset of student-submitted Python code, focusing on the detection of prevalent types of source code changes employed to conceal plagiarism attempts. With 28,701 pairwise comparisons across non-plagiarized and plagiarized code files, the model achieved an ac-curacy of 83%, with a high precision of 100%, a recall of 83%, and a 91% F-Measure. Ad-ditionally, the similarity detection process underwent a comprehensive assessment through test cases involving different degrees of source code transformation, demonstrating the model's robustness in identifying plagiarism across varying levels of structural and textual code transformations. Moreover, the K-Means algorithm exhibited high performance in grouping codes effectively, with optimal cohesion and separation validated using evalua-tion metrics such as the elbow method and silhouette score. Furthermore, a web application prototype was developed using Streamlit to operationalize the methodology. The applica-tion is user-friendly and provides an accessible platform where users can register, log in, and utilize the toolkit for source code similarity detection. The findings of the study under-score the effectiveness of integrating Abstract Syntax Tree (AST) analysis with a text-based approach, combined with the K-Means algorithm, as a scalable and comprehensive method for detecting source code similarity in Python programs.

Tags from this library: No tags from this library for this title. Log in to add tags.

Holdings
Item type	Current library	Shelving location	Call number	Copy number	Status	Date due	Barcode
Unpublished Materials	Main Library	Undergraduate Thesis Section	UTH BSCS D362p 2024 (Browse shelf(Opens below))	1-1	Not for loan		007022UM

Browsing Main Library shelves, Shelving location: Undergraduate Thesis Section Close shelf browser (Hides shelf browser)

Previous	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	No cover image available	Next
Previous	UTH BSCS C891e 2019 ESP32 based environment control and monitoring for poultry using fuzzy logic	UTH BSCS D127i 2024 Implementation of image recognition for automated campus parking /	UTH BSCS D239v 2023 Vehicle smoke emission detection using convolutional neural network /	UTH BSCS D362p 2024 Python code similarity detection using abstract syntax tree with k-means algorithm /	UTH BSCS D370p 2002 Proposed computerized class record system/	UTH BSCS D371m 2024 A mobile application for image-based characterization of medicinal plants using convolutional neural network /	UTH BSCS D3794a 2020 Applying speech recognition for basic windows command using VAD algorithm /	Next

Undergraduate Thesis (BSCS) - Computer Science. Camarines Sur Polytechnic Colleges, 2024.

Includes bibliographical references.

This study presents a hybrid methodology for detecting source code similarity in Python programming language. Integrating Abstract Syntax Tree (AST) analysis for structural similarity evaluation, the SimHash algorithm for textual similarity assessment, and the K-Means clustering algorithm for grouping Python codes according to their similarity fea-tures. The evaluation of the source code similarity detection was thoroughly conducted us-ing a dataset of student-submitted Python code, focusing on the detection of prevalent types of source code changes employed to conceal plagiarism attempts. With 28,701 pairwise comparisons across non-plagiarized and plagiarized code files, the model achieved an ac-curacy of 83%, with a high precision of 100%, a recall of 83%, and a 91% F-Measure. Ad-ditionally, the similarity detection process underwent a comprehensive assessment through test cases involving different degrees of source code transformation, demonstrating the model's robustness in identifying plagiarism across varying levels of structural and textual code transformations. Moreover, the K-Means algorithm exhibited high performance in grouping codes effectively, with optimal cohesion and separation validated using evalua-tion metrics such as the elbow method and silhouette score. Furthermore, a web application prototype was developed using Streamlit to operationalize the methodology. The applica-tion is user-friendly and provides an accessible platform where users can register, log in, and utilize the toolkit for source code similarity detection. The findings of the study under-score the effectiveness of integrating Abstract Syntax Tree (AST) analysis with a text-based approach, combined with the K-Means algorithm, as a scalable and comprehensive method for detecting source code similarity in Python programs.

There are no comments on this title.

to post a comment.