摘要
生成式人工智能(Generative Artificial Intelligence,GenAI)作为教育生态变革的重要驱动力,既加速了教育数字化的进程,也为数字化评估提供了更为科学、全面的新视角,推动了评估手段的革新。但随着大语言模型与多智能体技术的发展,如何利用其在知识提取、信息检索与协作博弈方面的优势,来应对指标体系构建繁杂、评估方式主观性强及评价周期冗长等痛点,成为区域教育数字化评估亟待解决的重要难题。为此,文章基于多智能体协作策略,提出了智能化的教育数字化评估方法,利用检索增强技术动态生成评估需求,并结合提示词工程技术引导多角色智能体协同完成评估指标体系设计与赋权。同时,针对多智能体的不同角色与功能,设计了同质协同、异质互补及多元整合三种多智能体协作模式,通过实践验证了该方法在区域教育数字化评估中的可行性与适切性。实验结果表明,该方法能够提升评估的精准度、效率与灵活性,为区域教育数字化发展的科学研判与决策优化提供了新的技术手段。
Generative artificial intelligence,as a crucial driving force for the transformation of the educational ecosystem,not only accelerates the process of education digitization but also provides a more scientific and comprehensive new perspective for digital assessment,promoting the innovation of evaluation methods.However,with the development of large language models and multi-agent technologies,how to leverage their strengths in knowledge extraction,information retrieval,and collaborative gaming to address the pain points such as complex indicator construction,strong subjective in evaluation methods,and long evaluation cycles has become an important and urgent problem to be solved in regional educational digital assessment.Therefore,based on the multi-agent collaboration strategy,this paper proposed an intelligent digital assessment method for education.It used retrieval enhancement technology to dynamically generate assessment requirements and combined prompt word engineering technology to guide multi-role agents to collaborate in completing the design and weighting of the assessment index system.Meanwhile,three collaboration models,namely homogeneous collaboration,heterogeneous complementarity,and multiple integration,were designed for the different roles and functions of multi-agents.The feasibility and appropriateness of this method in the digital assessment of regional education were verified through practice.The experimental results showed that this method can enhance the accuracy,efficiency,and flexibility of assessment,providing a new technical means for scientific judgment and decision-making optimization in the digital development of regional education.
作者
陈梦
陈旭
CHEN Meng CHEN Xu(Faculty of Artificial Intelligence in Education,Central China Normal University,Wuhan,Hubei,China 430079)
出处
《现代教育技术》
2026年第1期94-104,共11页
Modern Educational Technology
基金
教育部哲学社会科学研究重大项目“教育数字化转型的国际比较研究”(项目编号:22JZD045)
中央高校基本科研业务“高等教育教学智能评估关键技术研究”(项目编号:CCNU25JCPT021)资助。
关键词
多智能体
协作模式
检索增强生成
区域教育数字化评估
multi-agents
collaborative mode
retrieval augmented generation
regional educational digital assessment
