Sistem Pendeteksi Plagiat Pada Dokumen Teks Berbahasa Indonesia Menggunakan Metode Rouge-N, Rouge-L Dan Rouge-W

Abstract

Plagiarism is a serious problem in education. This research uses ROUGE-N (N = 3 or trigrams), ROUGE-L, and ROUGE-W (with weighted function f(x2)) at the sentence level to detect plagiarism on Indonesian language text documents. This research aim to obtain a suitable preprocessing for each method of assessment. The preprocessing includes stopword removal and stemming. This research uses clipping based on recall, precision, and f-measure. Analysis is restricted to preprocessing and calculation method used in each assessment method. Stemming improves ROUGE-N, while stopword removal negatively effects ROUGE-N. ROUGE-L and ROUGE-W performs well with f-measure clipping. ROUGE-W is better without stemming.