Multigranularity text classification and argumentative essay mining based on chain-of-thought distillation

Mining and analyzing arguments is an effective approach to mastering the writing methods of argumentative essays. In this paper, we propose a novel approach that integrates Chain of Thought (CoT) reasoning to enhance the performance of Large Language Models (LLMs). Specifically, we utilize GPT-4 to generate CoT data, enabling the model to develop a deeper comprehension of argumentative essay structures and improve its reasoning capabilities. Additionally, we introduce a two-stage fine-tuning framework tailored for both coarse-grained and fine-grained classification tasks. Experimental results on Pretrained Language Models (PLMs) and LLMs demonstrate significant enhancements in reasoning abilities, achieving substantial performance gains across all evaluation metrics. These findings underscore the potential of our method for advancing applications in argumentative understanding.