Sistem Pengukuran Tingkat Kemiripan Judul dan Abstrak Skripsi Menggunakan Algoritma Winnowing dan Dice Similarity

Thesis is a scientific paper that must be made by undergraduate stratum 1 academics at the UNHASY information technology faculty. One of the problems with writing a thesis at the Faculty of Information Technology, UNHASY, is that the process of checking the similarity of titles and abstracts is still manual so it takes a long time to check. From the existing problems, the research was carried out with the aim of providing a system for measuring the level of similarity of titles and abstracts of web-based thesis using the Winnowing and Dice Similarity Algorithm. The algorithm used in this research is the winnowing algorithm which is used as document fingerprinting. The title and abstract of the submitted thesis will be processed through several stages, namely Whitespace insensitive, forming an n-gram sequence, rolling hash, splitting the hash value into several windows and then taking the smallest fingerprint from each window. In the final stage, the similarity value is sought by using the dice coefficient. The results of this study are a sub-system that can check the similarity of titles and abstracts using the winnowing algorithm. This algorithm does not have specific provisions regarding how many gram, window, and prime values that must be applied so that the author tests using 10 test data to determine the effect of these values so that the smallest value is obtained from the input value gram = 5, window = 4 and prime = 3, which is 10.82%, while the largest result, which is 38.16%, is obtained from the input value of gram = 4, window = 3 and prime = 3.