Pendeteksian Kemiripan Kode Program C dengan Algoritme KMedoids

Abstract

Practice of plagiarism on program codes is more common and easier to do. Manual detection of plagiarism takes a lot of time and effort. The detection can be done by grouping program codes that have similar structures. This study intends to apply K-medoids algorithm on 4 C code program datasets to find similarities of code program and analyze clustering results. The experimental results show that the best clustering in dataset1 (If-Else Condition and Looping While) was obtained at k=10 with an average of dissimilarity 2.655, and 18.9% of students have the same group. In dataset2 (Looping While), the best clustering was obtained at k=9 with an average of dissimilarity 2.227, and 32.6% student assignments are in the same group. For dataset3, the assignments are divided into two clusters with an average of dissimilarty 0.719, and 87% of students assignments are in the same cluster. The best c lustering result on the dataset4 was obtained at k=6 with an average of dissimalrity 3.199, and 61% of students assignments are in the same group. The accuracy rate from clustering results is 93.28%