Automated Essay Scoring (AES) using English Essay Question

Abstract

Automated Essay Scoring (AES) is a one of the important research areas in educational technology. Research about AES is started in the early 1960s and keep growing in development along with the advances of the computing technology. AES is more widely use by the educational system for assessment and classroom activities, it being use in all layers of education. The potential for Automated Essay Scoring is being used widely becomes a reality as more and more classroom activities, assessment, and test preparation materials are supplied online. These essay questions might include everything from arithmetic, where students are asked to explain how they arrived at their answers, to science, where they might be asked to define words or explain experiments, or history, where they must ask to explain or discuss an event. In this study we try to design the AES using Natural Language Processing (NLP) and machine learning techniques to evaluate the student's answer by comparing it with the answer scheme. It is not simply a string-matching program and need a proper system to process the students answer. We propose the 4 main process to the system that is Spelling Checking, Pre-Processing, Latent Semantic Analysis and Thesaurus Checking. All this process is combined and analyze to become a good AES system. Each of this process mean give somethings important to the system flow on how the essay going to be evaluate. The system has been done some experiment and the system get the average correlation between system scores and human appraiser scores is only 70%. However, there needs to be more research on the future use of this Automatic Essay Scoring system.