【作 者】：
李 振， 周东岱
【关 键 词 】：
教育知识图谱； 概念模型； 知识元； 前驱后继关系； 认知状态； 学习路径
【栏 目】：
学习环境与资源
【中文摘要】：
自适应学习系统是实现个性化学习的重要突破口，而领域知识建模一直是困扰该系统发展的一大难题。目前，以深度学习、知识图谱为核心的新一代人工智能技术的回归，为其提供了新的发展契机。文章首先对已有的教育领域知识建模方法进行了梳理与总结，对其现存问题进行了对比分析；在此基础上，针对通用知识图谱迁移应用于教育领域所面临的知识粒度模糊、领域适应性不强两大问题，构建了一种教育知识图谱概念模型——EKGCM模型，该模型包括知识图示、认知图式两个层次，以及知识节点、知识关联、认知状态、学习路径四个基本要素；然后，针对图谱构建自动化程度不高的问题，文章提出一种基于智能处理技术的构建方法，具体包括知识元抽取、前驱后继关系挖掘、认知状态诊断、学习路径生成四个步骤；最后，采用理想智慧教育云平台中的教学资源和学习行为数据验证了上述方法的可行性。研究对于开展数据智能驱动的个性化自适应学习具有重要意义。
【英文摘要】：
Adaptive learning system is an important breakthrough to achieve personalized learning, while domain knowledge modeling is a major problem that has been puzzling the development of that system. At present, the new generation of artificial intelligence technology with deep learning and knowledge graph as the core provides a new opportunity for its development. Firstly, this paper summarizes the existing knowledge modeling methods in the field of education and analyzes the existing problems. On this basis, in view of two major problems faced by the application of general knowledge graph in the field of education, namely fuzzy knowledge granularity and weak domain adaptability, this paper constructs a conceptual model of educational knowledge graph - EKGCM model. That model includes two levels of knowledge representation and cognitive schema, and four basic elements of knowledge node, knowledge association, cognitive state and learning path. Then, in order to solve the problem of low automation of graph construction, this paper proposes a method based on intelligent processing technology, which includes four steps: knowledge element extraction, pre-and-after relationship mining, cognitive state diagnosis and learning path generation. Finally, the above methods are verified with the teaching resources and learning behavior data in an ideal cloud platform of smart education. The research is of great significance for the development of personalized adaptive learning driven by data intelligence.