Cost-based Framework for Natural Language Argumentation Analysis
Kazunori Yamaguchi, Yoshitatsu Matsuda · 2021
Classroom discussions are important for enriching the student's background knowledge and improving the student's logical thinking ability. Automatic evaluation of argument strengths in a discussion is helpful in such environments. In this paper, we propose a strength evaluation framework for such use. The evaluation framework is cost-based and allows attack and support relations with various semantics, recursiveness, and similarity. Its theoretical properties and application show that the proposed framework is promising.