DETEKSI PLAGIAT TESIS BERBAHASA INDONESIA MENGGUNAKAN METODE COSINE SIMILARITY

Abstract

Abstrack-Plagiarism detection plays a crucial role in maintaining academic integrity and intellectual honesty in higher education. One of the ways used to detect plagiarism is by using Cosine Similarity. Cosine Similarity works by comparing a document with other documents based on the number of keywords, bag of words and the frequency of occurrence of certain keyword words. However, the application of Cosine Similar ity is rarely analyzed for accuracy and performance. This research aims to test the performance of the Cosine Similar ity method and make a comparison with the Jaccard Similarity method. Data samples are obtained from Indonesian-language thesis data belonging to Budi Luhur campus students. The test model will be tested by comparing several original theses and documents containing plagiarism. The findings of this study show that Cosine Similarity has high accuracy in classifying documents as plagiarism or not, with an accuracy rate of 96.63%, showing its potential as an effective tool in detecting plagiarism. Meanwhile, the evaluation of Jaccard Similarity highlights the potential for improvement in the model's performance. This difference in accuracy is due to the difference in similarity measurement approaches between the two methods. Cosine Similarity measures similarity based on the direction of vectors in word space, while Jaccard Similarity calculates similarity based on the set of tokens. The results of this study make a significant contribution by testing and comparing the Cosine Similarity method with Jaccard Similarity in plagiarism detection. This research also provides a better understanding of plagiarism detection methods, but also offers new insights for the development of better models in detecting plagiarism. Finally, the results of this research affect policies or practices in educational institutions by providing a stronger basis for the enforcement of anti-plagiarism policies. It can encourage institutions to adopt technology-based plagiarism detection systems in the assessment process of theses and other academic works. In addition, the results may also help institutions raise awareness of the importance of academic integrity and encourage the development of more effective plagiarism prevention strategies in educational settings.