关键词: 自动作文评分, 评分原理要素, 自然语言处理, 认知思维能力

Abstract: This paper gives an overview of the principles, features and functions of the most well known automated essay scoring systems, including PEG, IEA, e rater and Criterion, IntelliMetric and MY Access!, and BETSY. Reliabilities of these systems are analyzed, and strengths and weaknesses of each of the systems are compared and contrasted. The paper analyzes the future development of AES systems on the basis of the discussion of former researches and the model of cognitively based assessment of writing competency. It summarize the orientation of development in seven aspects: design of AES systems on cognitively based assessment model of writing competence; shift of emphasis from surface features of grammar and structure to underlying features of critical thinking and rhetorical effect; expansion of subject areas, with focus on both English language arts and scientific reasoning; development of a new genre of AES software that can provide meaningful and effective formative feedback to assist writing process; use of machine learning technology to develop an open AES system that can address new problems automatically; enhancement of feedback effect via machine student dialogue; integration and cooperation of various disciplines of studies in the field of AES.