Automated essay scoring: A review of the field

Abstract

This paper critically reviews the recently published scientific literature on the task of Automated Essay Scoring (AES), by examining the various systems and approaches used. Automated Essay Scoring, which is the process of automating the evaluation of answers to open-ended questions, most usually in educational settings, by utilizing NLP techniques, gathers an increasing amount of interest, due to its potential applications both commercially and as part of the learn-ing process. The focus of this paper is to analyze the most popular approaches in recently published AES systems, categorize the systems with respect to certain characteristics in their design, the datasets that they use and the evaluation schemas that are used to evaluate them, and finally discuss the recent trends and challenges of the field of AES.