[2I5-J-9] Natural language processing, information retrieval: classification and evaluation

[2I5-J-9-05] Automatic Scoring and Scoring Support System using LSTM and Attention

〇Kohei Takai¹, Kengo Taketani¹, Junpei Hayakawa², Yasukuni Mori³, Hiroki Suyari³ (1. Department of Applied and Cognitive Informatics, Graduate School of Science and Engineering, Chiba University, 2. Department of Informatics and Imaging Systems, Faculty of Engineering, Chiba University, 3. Graduate School of Engineering, Chiba University)

Keywords:NLP, automatic scoring, scoring support, machine learning

In the university unified entrance examinations from 2020, a new kind of question to require answers in description style will be introduced. In this paper, we apply the scoring using LSTM and attention. We use about 1,200 score sheets and prepare two datasets((a), (b)) in the evaluation. (a) is randomly generated and (b) is generated using an automatic scoring system. As a result, the accuracy is 0.91 for (a) and 0.73 for (b).