A Latent Semantic Analysis Method for Automatic Scoring System at Essay Test

Abstract

An essay test is the type of test that requires students' skills to propose, arrange, and combine the ideas they have from the teaching-learning process conducted. The essay answers in the form of description cause the examination becomes more complicated compared to an objective multiple-choice test. Therefore, in this research, the scoring system is built for an essay test by Using Latent Semantic Analysis that can pay attention to the word synonym of the key answer on the students' answer. The highest accuracy from the test that had been conducted at the essay scoring system by Using Latent Semantic Analysis was 84.35%, categorized into good.