随着智慧教育的迅速发展,单一智能体系统在复杂教育环境中的适应性面临挑战。多智能体系统(MAS)通过智能体间的协作与交互,为个性化学习、协作学习和适应性教学提供了创新路径。文章系统梳理了教育智能体的内涵及其演进过程,提出了基于功能、过程和认知三维度的智能体分类框架,重点探讨了基于元认知理论、建构主义理论和自我调节学习模型的三种设计范式,并创新性地引入了Agcnt与LLM大语言模型融合的新范式。每种设计范式围绕不同学习过程中的计划、监控、反馈和协作等环节进行智能体功能划分,旨在提升学习者的自主学习能力,促进深度学习的实现。研究为多智能体系统在智慧教育中的设计与应用提供了系统化的理论支持,并展望了其跨场景协同和伦理治理的发展方向。
Abstract
With the rapid advancement of smart education, the adaptability of single-agent systems in complex educational environments faces signifcant challenges. Multi-agent Systems (MAS) provide innovative pathways for personalized learning, collaborative learning, and adaptive teaching through agent collaboration and interaction. This paper systematically reviews the connotation and evolution of educational agents, proposes a three-dimensional classifcation framework based on function, process, and cognition, and focuses on three design paradigms grounded in metacognitive theory, constructivist theory, and the Self-regulated Learning (SRL) model. Furthermore, it innovatively introduces a new paradigm integrating Agents with Large Language Models (LLMs). Each design paradigm delineates agent functions around various learning processes such as planning, monitoring, feedback, and collaboration, aiming to enhance learners' self-regulated learning capabilities and promote the achievement of deep learning The research provides systematic theoretical support for the design and application of Multi-agent Systems in smart education and outlines future directions, including cross-scenario collaboration and ethical governance.
关键词
多智能体 /
智慧教育 /
元认知 /
自我调节学习 /
大语言模型
Key words
multi-agent /
smart education /
metacognition /
self-regulated learning /
Large Language Model
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Raessens J, Goldstein J.Handbook of Computer Game Studies[M].Massachusetts:The MIT Press.2006
[2] Minsky M L.The Socicty of Mind[M].New York;Tokyo:Simon and Schuster,1986:19-32
[3] 李杨,徐峰,谢光强,等.多智能体技术发展及其应用综述J].计算机工程与应用,2018,54(9):13-21.
[4] 刘清堂,巴深,罗磊,等.教育智能体对认知学习的作用机制研究述评[J].远程教育杂志,2019,37(5):35-44.
[5] 郑娅峰,黄璟玥,包昊罡.教育智能体赋能科学教育:概念特征、应用价值与实施策略[J].远程教育杂志,2025,43(3):24-32
[6] 张静. AI智能体辅助混合式外语自主学习模式:设计、实施与优化[J].外语界,2025(3):29-37
[7] 王冬青,陈自力,邵文豪,等.从“负能”到“赋能”:基于LLMs的思维链提示设计与教研AI智能体构建一以课堂教学智能分析为例].中国电化教育,2025(3):111-117+125
[8] 郑娅峰,黄璟玥,包昊罡.教育智能体赋能科学教育:概念特征、应用价值与实施策略[J].远程教育杂志. 2025.43(3):24-32
[9] 吴永和,姜元昊,陈圆圆,等.大语言模型支持的多智能体:技术路径、教育应用与未来展望[J].开放教育研究,2024,30(5):63-75
[10] 顾小清,郝祥军.悟空的毫毛:正在重塑学习技术系统的多智能体[J].华东师范大学学报(教育科学版),2025,43(5):16-29.
[11] Outay FJabeur N,Bellalouna Fet alMulti-Agent System-Based Framework for an Intelligent Manmagement of Competency BuildingU].Smart Learning Environments, 2024,11(1):41.
[12] Viswanathan N,Meacham S,Adedoyin F F.Enhancement of Online Education System by Using a Multi-Agent Approach[J].Computers and Education: Artifcial Intelligence,2022,3:100057.
[13] 卢宇,余京蕾,陈鹏鹤.基于大模型的教学智能体构建与应用研究[J].中国电化教育,2024(7):99-108.
[14] Ciloglugil B,Alatli O,Inceoglu M M,et al A Multi-Agent Based Adaptive E-Learning System[C]// Computational Scicnce and Its Applications-ICCSA 2021.Cham:Springer, 2021:693-707
[15] 顾小清,郝祥军.悟空的毫毛;正在重塑学习技术系统的多智能体[J].华东师范大学学报(教育科学版),2025,43(5):16-29
[16] 黄昌勤,钟益华,王希哲,等.从单智能体到多智能体:大模型智能体支持下的激励型学习活动设计与实证研究[J].华东师范大学学报(教育科学版),2025,43(5):44-56.
[17] 朱俊华,许璐瑶,马近远.生成式人工智能如何赋能学生学习基于大学生自我调节学习行为的实证研究[J].高等工程教育研究,2025(2):66-72.
