During the past decades, researches about automatic grading have become an interesting issue. These studies focuses on how to make machines are able to help human on assessing students' learning outcomes. Automatic grading enables teachers to assess student's answers with more objective, consistent, and faster. Especially for essay model, it has two different types, i.e. long essay and short answer. Almost of the previous researches merely developed automatic essay grading (AEG) instead of automatic short answer grading (ASAG). This study aims to assess the sentence similarity of short answer to the questions and answers in Indonesian without any language semantic's tool. This research uses pre-processing steps consisting of case folding, tokenization, stemming, and stopword removal. The proposed approach is a scoring rubric obtained by measuring the similarity of sentences using the stringbased similarity methods and the keyword matching process. The dataset used in this study consists of 7 questions, 34 alternative reference answers and 224 student's answers. The experiment results show that the proposed approach is able to achieve a correlation value between 0.65419 up to 0.66383 at Pearson's correlation, with Mean Absolute Error (𝑀𝐴𝐸) value about 0.94994 until 1.24295. The proposed approach also leverages the correlation value and decreases the error value in each method.

Abstract. Assessment is one of the most important things in studying. In this digital era, there are a lot of systems that have developed to handle assessment automatically. One of the system assessments that were developed by researcher is automatic scoring for essay. There are two types of essay; long and short answer essays. This paper is focused on the development of automatic short answer scoring. Some automatic scoring systems used on long answer have shown optimal results in giving a score on the students answer. Automatic long answer systems use the information retrieval method to measure similarity between students answer and references answer. Automatic short answer scoring does not give the best result yet. Short answer has a limited word in each answer. Each answer consists of one phrase to three sentences. Assessment of the short description that has limited number of words requires special handling, especially in the weighting process. With the limitations of the process of weighting the word, it cannot be done with frequency model, because the words occurrence is very rare. This study tries to compare several methods that apply the overlapping methods to determine the degree of similarity between the references answer and students answer. From the research result shows that the method Cosine Coefficient has better results than the Dice and Jaccard Coefficient methods.

No abstract

No abstract