Development of Paragraph Automated Essay Scoring System of Chinese Writing

At present, most of the automated essay scoring systems used in teaching score the whole essay. Therefore, we develop an automated essay scoring system that scores and comments instantly on each paragraph. The feedback supports elementary students to practice writing essays and improve their writing skills. The study also researches the system’s usability. This developed writing essay scoring system uses training models created with Python. Firstly, the collected 166 essays are classified into each paragraph, then data is classified into 1−6 grades according to the composition scoring rules set by the experienced teacher. Secondly, training data performs feature extraction by ALBERT, a lite BERT. Then, training data is used to train the deep learning of the scoring model, the Bi-GRU model. Thirdly, the scoring model is linked to the comment database. The comment is selected according to the scoring model’s result. Finally, the models are combined to complete the automated essay-scoring system. When the user input content of the paragraph, the system links it to the scoring model and comment database of the paragraph. After the scoring model’s judgment is completed, the score and the comment feedback are output. The user can modify the content according to the comment until the modification is completed. The user continues to write until the completion of the essay. Elementary school teachers tested the system developed, received interviews, and filled out questionnaires to understand the usability of the system and its impact on learners' writing learning. Elementary school teachers scored an average of 87.9 points on the system developed in this study. The system was commented on for good impressions and ease to use. They also agree with the practicality of teaching. In scoring, the teachers had different results.