多模态学习分析:走向计算教育时代的学习分析学  

Multi-modal Learning Analytics:Learning Analytics Towards the Age of Computational Education

作　　者：张琪 李福华[1] 孙基男[2] Zhang Qi;Li Fuhua;Sun Jinan(College of Education,Huaibei Normal University,Huaibei 235000,Anhui;National Engineering Research Center for Software Engineering,Peking University,Beijing 100871)

机构地区：[1]淮北师范大学教育学院,安徽淮北235000 [2]北京大学软件工程国家工程研究中心,北京100871

出　　处：《中国电化教育》2020年第9期7-14,39,共9页China Educational Technology

基　　金：国家社会科学基金教育学一般课题“多维视角下人工智能教育应用的理论框架和实践路径研究”(课题编号:BCA190082)阶段性研究成果。

摘　　要：利用跨数据源分析教与学规律的研究已成为教育技术学演进的重要力量。基于对多模态学习分析本质的理解,构建冰山隐喻分析模型,以阐释多模态数据流、数据建模与学习状态输出之间的关系,进而构造循环推理框架。该框架以环状流动表征了多模态学习分析的流程,流程起始于涵盖学习行为的输入空间,通过数据空间的分配标注,并在同步空间进行时空匹配使数据得以对齐,进而在融合空间中实现对“数字-推理”区域的构造,最终实现反馈输出以及对学习行为进行引导与干预的目的。多模态学习分析的研究需要继续加强复杂计算模型构建,建立学术研究共同体与开源生物数据库,扩展当前的互操作性标准,加强共同学习的模态互补研究,并将伦理价值和准则内嵌于框架设计之中,从而支撑计算教育学的建立与下一代互联网教育创新。The research,applying cross-data sources to analyze learning processes,has become an important part in the evolution of educational technology.Based on the understanding of the nature of multi-modal learning analytics,an iceberg metaphor analytic model is constructed.This framework uses a circular flow to characterize the process of multi-modal learning analytics.The process starts from the input space covering the learning behavior and assign labels to raw data in data space.Then,it performs space-time matching to align the data in the synchronization space and the structure of the“digital-inference”area is realized in the fusion space.In the end,the feedback output and goals of the guidance and intervention to learning behavior are achieved.Research on multi-modal learning analytics needs to continue to strengthen the construction of complex computing models,establishing academic research communities and open source biological databases,expanding current interoperability standards,strengthening modal complementary research for common learning,and embeds ethical values and guidelines in the framework of the design in order to support the establishment of computational education and the innovation of the next generation of Internet education.

关 键 词：多模态 学习分析 计算教育学 数据映射 研究框架 

分 类 号：G434[文化科学—教育学]