Automatic Essay Assessment for Blended Learning in Elementary School

Abstract

Blended learning combines traditional face-to-face learning in class and virtual learning. This system requires an evaluation process as a measuring instrument. This paper develops an automated essay correction on a blended learning system for elementary school. Assessment in education is obtaining, organizing, and presenting information about what and how the student learning. Open-answer questions allow teachers to understand the student's answer. Essay questions can be used to train students in conveying information verbally and measure their understanding. The teacher needs more time to examine the essay answers for each student. The essay correction needs to be guided with a scoring rubric as the keyword in the answer key that automatically makes essay corrections for elementary school. This system uses the Rabin Karp method to measure the similarity between answer keys to students' answers. The test was carried out by comparing Mean Absolute Error and Pearson Correlations from various k-gram values. The experiments show this assessment system produces a small error value and good performance in grading the student's answer with a low difference value between automatic assessment and expert judgment. Further research, this system can be applied to evaluate the student learning outcomes in an integrated manner with STEAM elements through blended learning. The use of automatic essay assessment in blended learning can improve elementary school students writing skills in the digital educational environment 4.0.