Pendeteksi Kemiripan Teks Paragraf dalam Dokumen Menggunakan Algoritma Leacock Chodorow dan Cosine Similarity

Abstract

Plagiarism is the act of closely copying or imitating, taking works from authors or creators without prior permission, with the intention of claiming the work as one's own original creation. Plagiarism is a common issue in both general and academic settings. In the academic context, plagiarism poses a significant problem because verifying the authenticity of a document is time-consuming and varies in terms of precision. Therefore, an approach need to be to developed aimed at detecting paragraph similarity in documents. This research aims to address the issue by automatically detecting similarity in the content of scholarly documents based on paragraphs, using the Leacock Chodorow algorithm and cosine similarity. In the system testing, 28 test data and reference data are used, with 7 test documents consisting of a total of 177 test paragraphs and 21 reference documents consisting of a total of 438 reference paragraphs. The evaluation results of the paragraph similarity detection system obtained an accuracy of 0.923 or 92.3%, precision of 0.908 or 90.8%, recall of 0.953 or 95.3%, and an F-measure of 0.930 or 93%.