多模态学习分析:“多模态”驱动的智能教育研究新趋向

Multimodal Learning Analytics:a New Research Trend in Intelligence Education Driven by Multimodality

智能感知技术的发展为学习分析领域的变革提供了多元的数据支持,实现智能技术与教育研究的深度融合,推动多模态学习分析的研究进展。该文立足于人工智能时代“多模态”的价值内涵,从多模态信息感知、多模态数据表征、多模态计算方法三个方面实现了“多模态”意蕴的多元分解。论述了多模态学习分析的历史沿革和研究意义,从多源性、层次性、时序性、情境性四个层面分析了多模态学习分析的潜在特征。人工智能时代多模态学习分析研究的开展需要从面向智能精准测评的学习者建模、面向智慧学习空间的教育情境建模、面向教育生态治理的教育系统建模三个方面加以重视,实现多模态学习分析与真实教育场景的深度融合。未来多模态学习分析研究的开展需要从多学科融合的基础理论体系建构、多模态数据集的创建共享、学生认知结构的量化表征、基于多模态学习分析的智能服务体系以及数据安全和伦理道德规范等方面加以重视。

The development of intelligent perception technology provides multiple data support for the reform of learning analysis field,realizes the deep integration of intelligent technology and education research,and promotes the research progress of multimodal learning analytics.Based on the value connotation of“multi-modality”in the era of artificial intelligence,this article realizes the multi-decomposition of“multi-modality”meaning from three aspects:multimodal information perception,multimodal data representation,and multimodal calculation methods.This paper discusses the historical evolution and research significance of multimodal learning analytics,and analyzes the potential characteristics of multimodal learning analytics from four levels:multisource,hierarchical,sequential,and situational.The development of multimodal learning analytics research in the era of artificial intelligence needs to pay attention to three aspects:learner modeling for intelligent and accurate evaluation,educational context modeling for smart learning space,and educational system modeling for educational ecological governance,so as to strengthen the deep integration of multimodal learning analytics and real education scenarios.In the future,the development of multimodal learning analytics research needs to pay much attention to the following aspects:the construction of the basic theoretical system of multidisciplinary integration,the creation and sharing of multimodal datasets,the quantitative representation of students’cognitive structure,the intelligent service system supported by multimodal learning analytics and the problem of data security and ethics based on multimodal learning analytics.