【作 者】：
汪维富， 毛美娟
【关 键 词 】：
多模态学习分析； 物理学习环境； 可穿戴设备； 生理数据； 数据融合； 学习评价； 认知带
【栏 目】：
理论探讨
【中文摘要】：
随着物联网、可穿戴传感设备、人工智能的快速发展，多种模态数据的连续抓取与融合分析成为可能，促成了多模态学习分析研究的诞生，并成为学习分析领域的一个重要新兴分支。多模态学习分析关注真实的学习情境与本真的学习交流方式，“返璞归真”地聚焦更普遍的物理学习情境，克服了主流学习分析过度关注数字化学习环境的局限。研究主要采用文献归纳法，阐述了多模态学习分析的发展动因、概念理解，重点从身体、生理视角概述了多模态的数据类型，并对多模态学习分析的过程模型进行了详细描述。研究认为，未来要充分借助人工智能处理多模态数据，借鉴认知带理论融合不同模态数据的意义，借力学习理论来理解真实的多模态学习交互过程，以进一步完善基于多模态数据的学习评价方法。
【英文摘要】：
With the rapid development of the Internet of Things, wearable sensing devices and artificial intelligence, continuous capture and fusion analysis of multimodal data has become possible, leading to the birth of multimodal learning analytics research, which has become an important emerging branch of learning analytics. Multimodal learning analytics focuses on real learning contexts and authentic learning communication modes, and focuses on more general physical learning contexts, overcoming the limitations of the mainstream learning analytics that pays too much attention to digital learning environments. The research mainly uses the literature induction method, expounds the motivations for the development of multimodal learning analytics, its conceptual understanding, with a focus on physical and physiological overviews of multimodal data types, as well as a detailed description of the process model for multimodal learning analytics. The study concludes that in the future, it is important to fully leverage artificial intelligence to process multimodal data, draw on cognitive band theory to fuse the meanings of different modal data, and use learning theory to understand the real multimodal learning interaction process in order to further refine the multimodal data-based learning evaluation methods.