Deep-neural Automated Essay Scoring: A Review

Abstract:
Automated essay scoring (AES) is the task of automatically assigning scores to essays as an alternative to manual scoring. Traditional AES models typically rely on handcrafted features, whereas deep neural networks (DNNs) -based AES models use automatic feature selection from the original texts. Furthermore, DNNs -AES models have recently achieved fantastic results and attracted increased attention. To our knowledge, various DNNs-AES models have been designed over the past few years, but no study has reviewed DNNs -AES models from the development process of NLP techniques. Therefore, this paper classifies DNNs-AES models into two categories and introduces existing representative DNN-AES models according to this classification. The main idea and detailed architecture of each model are described.