Enhancing Regional Plagiarism Detection Using a Backtrack Matching Model: A Precision, Recall, and F1 Score-Based Evaluation

Abstract

In the era of digital content creation and academic publishing, ensuring the originality of written work has become crucial. Plagiarism detection tools have evolved to combat this issue, but their effectiveness varies based on regional linguistic nuances, such as local idioms, translations, and cultural expressions. This paper presents an evaluation framework for a regional plagiarism checker (Using Hindi, Marathi, Tamil, Gujarati, etc.…), assessing its performance using key metrics: precision, recall, and F1 score. Precision measures the accuracy of identified plagiarism instances, while recall evaluates the tool’s ability to detect all actual instances of plagiarism. The F1 score provides a harmonic meaning, balancing both precision and recall. The evaluation highlights the challenges of detecting plagiarism in regional contexts and comparison, where language variations can lead to detect or undetected content from similarity. Our results show that by tailoring plagiarism detection algorithms to regional linguistic characteristics, it is possible to significantly improve detection accuracy and reduce false positives, thus providing a more effective and reliable tool for regional academic integrity