Perbandingan Algoritma Winnowing dan Algoritma Rabin-Karp pada Aplikasi Pendeteksi Kesamaan Dokumen Skripsi

Abstract

Plagiarism is often found in the academic world and is considered a serious violation because it involves the appropriation of ideas, opinions, or writings of others. A thesis is the final work of a student that must meet scientific standards and be tailored to the field of study. It is important to check the similarity of the thesis using a web-based sistem to prevent plagiarism and ensure academic integrity. This sistem will be developed as a web-based platform with the aim of comparing the Winnowing algorithm and the Rabin-Karp algorithm in checking the similarity of thesis/final project texts with existing thesis data. In principle, both methods involve searching for strings using hashing functions to compare the sought string (m) with the compared string (n) by comparing the results of the hashing function used. However, the Winnowing algorithm differs in that it does not use all hash values from each formed set of grams. The hash values formed in the previous stage will be divided into a window of size (w). In this research, the sistem testing uses data from Computer Engineering students at Halu Oleo University to facilitate checking the plagiarism level of theses using the Rabin-Karp and Winnowing algorithms. In this study, the Rabin-Karp and Winnowing algorithms have been implemented successfully in the plagiarism checking sistem for students' theses. The test results for the comparison of the Winnowing and Rabin-Karp algorithms in terms of processing time show that the Rabin-Karp algorithm takes 1.509 seconds, while the Winnowing algorithm takes 1.508 seconds. Subsequent testing using Normalized Mean Absolute Error (NMAE) reveals that the Rabin-Karp algorithm has an absolute error value of 0.1829, while the Winnowing algorithm has a value of 0.0194. Therefore, based on the NMAE test, the Winnowing algorithm performs better than the Rabin-Karp algorithm.