ACTOR: Advancing Argument Components Identification Through In-Context Learning and Proximity Information Awareness

Peijian Zeng, Weixiong Zheng, Hongyan Wu, Nankai Lin, Aimin Yang, Shengyi Jiang

Abstract: Argumentative writing plays a crucial role in developing and enhancing high school students’ critical thinking skills, serving as a key form of argumentative expression. The task of identifying argument components aids students in understanding the structure of argumentative essays and assists teachers in evaluating students’ proficiency in scientific argument mining. However, existing research lacks a detailed classification of argument types. This paper addresses the NLPCC 2024 Shared Task 5 “Argument Mining for Chinese Argumentative Essays” by proposing a Argument Components idenTificatiOn fRamework (ACTOR). We employ a proximity information awareness (PIA) strategy to provide the model with more relevant information and use the in-context learning (ICL) method to offer pertinent reference examples. Experimental results indicate that our method is competitive in the argument component identification task.