Development of an automated essay scoring system over the web

Abstract

This report narrates the development of Automated Essay Scoring (AES) system over the web. AES is the use of specialized computer programs to recognize and grades to written essays in an educational setting. The criteria for an AES to be recognized as a valid system are the method of assessment must be judged on validity, fairness, and reliability. There are many different approaches to design an AES model, however among the variety of methods, there is an approach, namely Gated Convolutional Neural Network (Gated-CNN) [1] which incorporates both character and word information into neural-network model that outperform the other state-of-the-art models on the Kaggle’s Automated Student Assessment Prize (ASAP) dataset [2]. However, marking an essay usually consists of different kinds of aspects, such as grammar usage, content relatedness, vocabulary spelling and others. Thus, it is essential to investigate the evaluation criteria of Gated-CNN. Experiments conducted and results are reported in the early part of this report. The details about the web AES system implementation are reported in the latter part of this report. It describes the perspective, architecture, and features of the web AES system.