【摘要】 作为面向教育场景的定制化人工智能模型，教育大模型在实现个性化发展、提升教学质量、促进教育公平等方面展现出了巨大的潜力。然而，教育大模型的发展面临数据质量、算法安全、伦理责任等多方面的挑战，且存在底层架构不统一、教育适配性不足等问题，亟需相关标准体系的引导与支撑。基于此，文章聚焦于世界数字教育联盟发布的首项标准——《教育大模型：总体参考框架》，首先介绍了其内涵、现状和挑战，并阐明了其理论基础与设计原则，以厘清此框架研制的构建逻辑；之后，文章阐释了其核心内容，指出此框架采用五层架构（自底向上依次为基础层、数据层、模型层、接口层、应用层），并包含“安全、伦理与隐私”“治理”两个核心要素。教育大模型总体参考框架标准的研制，为教育大模型的建设提供了标准化支撑，有助于其实现科学、有序、可信发展，进而推动教育数字化转型和高质量发展。更多还原

【Abstract】 As customized artificial intelligence models faced for educational scenarios, large model for education demonstrate significant potential in achieving personalized development, improving teaching quality, and promoting educational equity. However, the development of large model for education is confronted with challenges including data quality, algorithmic safety, and ethical responsibility, alongside issues such as inconsistent underlying architecture and insufficient educational adaptation, which urgently requires guidance and support from a relevant standards system. Based on this, the paper focused on the first standard released by the World Digital Education Alliance — Large Model for Education: Overall Reference Framework. This paper first introduced its connotation, status, challenges, and clarified its theoretical foundations and design principles to elucidate the construction logic of the framework development. Subsequently, this paper elaborated on its core content, pointing out that the framework adopted a five-layer architecture(from bottom to top: infrastructure layer, data layer, model layer, interface layer, and application layer) and incorporated two key elements: “security, ethics and privacy” and “governance”. The development of the standard for overall reference framework of large model for education provided standardized support for the construction of large model for education, contributing to their scientific, orderly, and trustworthy development, and further promoting the digital transformation and high-quality development of education.更多还原