Automatic Essay Grading for Bahasa Indonesia with Support Vector Machine and Latent Semantic Analysis

Abstract

This research is used to increase the accuracy of automatic short essay grading for Bahasa Indonesia. Short essay in Bahasa Indonesia are classified using Support Vector Machine (SVM) based on its topic to decrease unrelated answer then assessed using Latent Semantic Analysis. Term Frequency-Inverse Document Frequency (TF-IDF) is used to weigh word in short essay and the result will be an input on SVM. The output of SVM are assessed using Latent Semantic Analysis (LSA). Latent Semantic Analysis uses Term Frequency Matrix to represent text in matrix, Singular Value Decomposition to decompose these matrix, and Frobenius Norm to find the similarity of lectures' answer and students' answer In this research, parameter C value as 1 and kernel linear are used to obtain the highest accuracy of classification using Support Vector Machine, 97,297% with 50% portion of data as training and 50% portion of data as testing. The accuracy score obtained from LSA is 72,01%.