摘要
面对新时代高质量教育与创新人才的发展需求,传统学习评价的弊端日益凸显,亟需创新体制、深化改革。多模态学习分析通过多维度全面采集和分析学习过程动态数据,使面向过程与真实学习情境的学习评价成为可能。文章首先基于当前学习评价的难点问题分析,剖析了多模态学习分析应用于学习评价的价值内蕴。随后,文章提出四类多模态数据所映射的评价表征维度和内容,从确立目标、获取数据、建构模型、提供反馈四个方面构建并阐释了基于多模态学习分析的学习评价流程。最后,文章结合相关研究成果提出多模态学习分析的学习评价应用建议,以期为面向过程的发展性学习评价改革赋能助力,并为高阶学习评价提供新视角。
Faced with the development needs of high-quality education and innovative talents in the new era,the drawbacks of traditional learning evaluation have become increasingly prominent,and it is urgent to innovate the system and deepen reform.Multimodal learning analysis made learning evaluation oriented to process and real learning situations possible through comprehensively collecting and analyzing the dynamic data of the learning process from multiple dimensions.Firstly,based on the difficult problems analysis of current learning evaluation,this paper analyzed the intrinsic value of multimodal learning analysis applied to learning evaluation.Subsequently,this paper proposed the dimensions and contents of evaluation representations mapped by four types of multimodal data,constructed and explained the learning evaluation process based on multimodal learning analysis from four aspects of establishing goals,acquiring data,constructing models and providing feedback.Finally,combined with relevant research results,this paper put forward application suggestions of the learning evaluation of multimodal learning analysis,expecting to empower the process-oriented developmental learning evaluation reform and provide a new perspective for the higher-level learning evaluation.
作者
张家华
胡惠芝
黄昌勤
ZHANG Jia-hua;HU Hui-zhi;HUANG Chang-qin(Key Laboratory of Intelligent Education Technology and Application of Zhejiang Province,Zhejiang Normal University,Jinhua,Zhejiang,China 321004)
出处
《现代教育技术》
CSSCI
2022年第9期38-45,共8页
Modern Educational Technology
基金
浙江省哲学社会科学规划课题“新一代人工智能支持下课堂教学改革方案构建与策略优化”(项目编号:22YJRC02ZD-2YB)
浙江师范大学教师教育学院开放基金项目“数据驱动的师范生智慧教学能力评测体系构建与应用”(项目编号:jykf22003)的阶段性研究成果。
关键词
多模态数据
学习分析
学习评价
学习过程
数据表征
multimodal data
learning analysis
learning evaluation
learning process
data representation