Plagiarism detection using Manber and Winnowing Algorithm

Plagiarism is a copy of the essay or opinion of others and does not list the written references, and makes it his own essay or opinion. There are several algorithms that have the ability to detect plagiarism of documents such as Jaro-Winkler algorithm, winnowing, Manber and others. In this study, research conducted on Mamber and Winnowing algorithms in detecting plagiarism. The Manber algorithm is an algorithm that uses K-grams but does not use the formation of a window while the winnowing algorithm is an algorithm that uses the K-grams approach in shaping the fingerprint pool. The app divides the documents into Biword and Triword tokens. These tokens are converted to MD5 value, the tokens have a hash value that has the same length and can be used as a document fingerprint. The Biword and Triword approaches are implanted in the winnowing algorithm, while the Biword is for Manber algorithms. This algorithm can check the phrase of each document, then saved in to an array. At the time of displaying the document will be obtained the same value long, the algorithm is able to display the value of arrays that form a Biword token as a fingerprint. From the results of the similirity of the 10 test data, the average result for manber algorithm is 90.56%, the Winnowing algorithm is 94% and the Winnowing triword 91.22% algorithm. The average time of generating winnowing triword data is 78.95 seconds and is 5.2% slower than the winnowing biword of 73.75 seconds