Presenter:Maocheng LIANG
Automated Essay Scoring with Artificial Neural Network
Maocheng LIANG
Beijing Foreign Studies University
Research on Automated Essay Scoring (AES) began in the 1960's (Page 1968), when Ellis Page extracted a number of textual features from student essays and developed a generalized linear model (GLM) to predict essay scores. However, it was not until the late 1990's and the beginning of the new century that AES attracted more scholarly attention (Shermis & Burstein, 2003). Since then, several systems have been developed, some of them now being operational, including ETS's e-rater. Most of these systems, like Page's, predict essay scores with a GLM, a statistical approach which assumes that learners' writing proficiency develops linearly. Contrary to this assumption, many researchers (e.g., Larsen-Freeman, 1997) point out that this is not the case. This study reports our work with a technique in Deep Learning, namely, the Artificial Neural Network (ANN). Findings of the study indicate that ANN models generate more reliable scores.
Key Words: Automated Essay Scoring, Artificial Neural Networks, score reliability
Page, E.B. (1968). The Use of the Computer in Analyzing Student Essays. International Review of Education 14(3): 253-263.
Shermis, Mark D. & Jill C. Burstein (eds). (2003). Automated Essay Scoring: A Cross-Disciplinary Perspective. Mahwah, NJ: Lawrence Erlbaum Associates.