Improvement of Argumentation Structure Visualization for Feedback on Students' Reports

This study focuses on developing an effective argumentation structure visualization system for helping teachers in reviewing students' reports. In university education, students have opportunities to join logical writing classes with exercises. In addition, they have more opportunities to receive feedback to update their arguments in their reports in many other classes outside of the logical writing class. However, students often struggle to update their arguments in the reports although they have received the feedback about the inappropriate points of the argument from teachers. Moreover, such students sometime could not understand the feedback itself. One of reasons for such problems is that there are some differences between the argumentation structures the student has tried to write in the report and those that teachers can read from the report. Therefore, we believe that teachers' feedback must have a figure for the argumentation structures that teachers can read from the report. To reduce the cost to create feedback with the figure by teachers, we develop an argumentation structure visualization system for supporting teachers to generate such a figure that the teachers can read from the report. In this study, we improve the system's performance of argumentation structure generation from the students' report for applying the system into the practical environment. We inserted BERT into the analyzing components of the argument component classification and the argumentative relation identification. We also evaluated the extended system as reducing the teachers' cost for their feedback creation.