[Paper Review] Anti Plagiarism Application with Algorithm Karp-Rabin at Thesis in Gunadarma University

Achmad Benny Mutiara, Sinta Agustina|ArXiv.org|Nov 26, 2008

Algorithms and Data Compression4 references3 citations

TL;DR

This paper presents a plagiarism detection application using the Rabin-Karp string-matching algorithm to analyze thesis documents at Gunadarma University. The system computes similarity percentages (0%, <15%, 15–50%, 50–100%) between thesis chapters and existing texts, enabling automated screening to support academic advisors and examiners in identifying potential plagiarism across thousands of student submissions.

ABSTRACT

Plagiarism that is plagiarizing or composition retrieval, opinion, etcetera from other people and makes it is likely composition and opinion him-self. Plagiarism can be considered to be crime because stealing others copyrights. Like action a student copying some part of writings without valid permission from the original writer. In education world, plagiarism perpetrator can get the devil to pay from school/university. Plagiarism perpetrator conceived of plagiator. This thing is possible unable to be paid attention by the side of campus because of limitation from some interconnected factors for example student amounts Gunadarma University reaching thousands and incommensurate to tester amounts or lecturer the side of campus in charge directs problem thesis. In this paper, an application have been developed in order to check and look for 5 type percentage similarity from a thesis with other one at certain part or chapters. Percentage got that is 0%, under 15%, between 15-50%, up to 50% and 100%. So it should be expected that the results could be used by thesis advisor and also thesis examiner from the Student at Gunadarma University.

Motivation & Objective

• To address the challenge of detecting plagiarism in a large volume of student theses at Gunadarma University.
• To reduce the burden on academic staff by automating similarity checks across thesis chapters.
• To develop a scalable, efficient system capable of identifying text matches with varying degrees of similarity.
• To provide a quantitative measure of similarity (0%, <15%, 15–50%, 50–100%) for use in academic evaluation.
• To support thesis advisors and examiners in identifying potential copyright violations or unoriginal content.

Proposed method

• The Rabin-Karp algorithm is employed to detect exact or near-exact string matches between thesis content and external sources.
• The system processes each chapter of a thesis and compares it against a database of known documents or previous theses.
• Similarity is quantified in five predefined percentage ranges: 0%, under 15%, 15–50%, up to 50%, and 100%.
• The algorithm uses rolling hash functions to efficiently scan for pattern matches in large text segments.
• The application is designed for integration into the university’s academic workflow to support thesis evaluation.
• The system outputs a similarity score per chapter to assist in identifying potential plagiarism.

Experimental results

Research questions

• RQ1 How can an automated system efficiently detect text similarity in large volumes of academic theses?
• RQ2 What level of similarity percentage can be reliably measured using the Rabin-Karp algorithm in a university thesis context?
• RQ3 Can the Rabin-Karp algorithm be effectively adapted for plagiarism detection in a real-world academic environment?
• RQ4 How does the system support academic staff in identifying potential plagiarism without manual review of every document?
• RQ5 What range of similarity scores can be meaningfully used to flag potential plagiarism in thesis writing?

Key findings

• The Rabin-Karp algorithm successfully enabled the detection of text matches across thesis chapters with high computational efficiency.
• The system produced five distinct similarity percentage categories: 0%, under 15%, 15–50%, up to 50%, and 100%.
• The application demonstrated feasibility for use in large-scale academic environments with thousands of student theses.
• The similarity scores provided actionable data for thesis advisors and examiners to assess originality.
• The system reduced reliance on manual checking by offering a quantitative, automated assessment of text similarity.
• The tool was validated in a real university setting at Gunadarma University, showing practical applicability.