摘要
论证式协作知识建构是协作知识建构的重要模式,能有效促进学生知识创新、论证技能习得和高阶思维发展。然而,论证式协作知识建构教学常因学生论证技能欠缺、协作效果不佳、规避认知冲突等问题而难以充分发挥其潜力。已有研究采用教学脚本、协作支架和对话代理等干预加以应对,但存在抑制自然交互、灵活性不足、难以应对复杂教育场景等问题。以大语言模型为核心的多智能体技术因其具备自然交互、动态分析和复杂语义理解能力,为解决上述问题提供了可能。基于此,本研究开发了一个基于大语言模型的多智能体系统ABCKC-AI,从参与、论证、认知和社会共建四个维度设计了自适应的协作动态支持,并开展实验研究以探索系统的有效性。研究结果表明,ABCKC-AI的动态协作支持显著提升了学习者的论证能力和协作成果质量,拓展了讨论的广度和深度,提高了论证结构和知识建构层级;其易用性与有用性也获得了学习者的高度认可。最后,本文讨论了ABCKC-AI作用于论证式协作知识建构的影响机制、在教育教学中的实践启示,以及该研究领域的未来推进方向,为大语言模型与协作知识建构的深度融合提供了创新思路、技术框架与实证依据。
Argumentation-based collaborative knowledge construction(ABCKC)is an important mode of collaborative knowledge construction that promotes students’knowledge innovation,acquisition of argumentation skills,and development of higher-order thinking.However,ABCKC often fails to reach its full potential due to students’lack of argumentation skills,ineffective collaboration,and the avoidance of cognitive conflict.While prior interventions,such as teaching scripts,collaborative scaffolds,and conversational agents,have sought to address these issues,they often inhibit natural interaction,lack flexibility,and are ill-suited to complex educational settings.Multi-agent technologies powered by large language models(LLMs)offer a promising solution by enabling natural interaction,real-time analysis,and complex semantic understanding.Accordingly,this study designed ABCKC-AI,an LLM-based multi-agent system that provides adaptive,dynamic support across four key dimensions of ABCKC process,including participation,argumentation,cognition as well as social co-construction,and conducted an experiment to evaluate its effectiveness.Results indicate that ABCKC-AI significantly improved learners’argumentation skills and the quality of collaborative products,broadened the scope and depth of discussion,strengthened argumentation structures,and deepened levels of knowledge construction.Learners also rated the system highly for usability and usefulness.Finally,the paper discusses the mechanisms through which ABCKC-AI shapes ABCKC process,the pedagogical implications,and directions for future research.It contributes to the field of LLM-supported collaborative knowledge construction by offering innovative insights,a technical framework,and empirical evidence.
作者
欧阳璠
付宏杰
Ouyang Fan;Fu Hongjie(College of Education,Zhejiang University,Hangzhou Zhejiang 310058)
出处
《远程教育杂志》
北大核心
2025年第5期41-54,共14页
Journal of Distance Education
基金
2022年国家自然科学基金面上项目“在线协作学习中群体认知发展机制研究:计算建模、分析反馈及教学干预”(项目编号:62177041)的研究成果。
关键词
人工智能教育
生成式人工智能
多智能体
论证式协作知识建构
协作知识建构
大语言模型
Artificial intelligence in education
Generative artificial intelligence
Multi-agent system
Argumentation-based collaborative knowledge construction
Collaborative knowledge construction
Large language models
