摘要
针对教师课堂提问方式量化模型四何问题的粒度过粗与耗时费力的缺陷,借鉴布鲁姆教育目标分类理论,设计了一个细粒度四何问题模型,可同时从问题类型和问题倾向两个维度对教师课堂提问方式进行观测.此外,还引入了深度学习模型长短期记忆神经网络LSTM,实现教师课堂提问的细粒度自动分类.实验结果表明,该方法能够高质量地实现问题分类(四何问题准确率达95%,细粒度四何问题准确率达84%),有效实现了课堂观察数据的智能整理.
In view of the limitations such as prohibitive cost and unreasonable question granularity raised by 4 MAT question model,a fine-grained question classification model is proposed based on Bloom’s classification theory of educational objectives.Through the proposed model,teachers’classroom questioning can be synchronously understood from the perspectives of question types and thinking dispositions.Moreover,the classical deep learning model LSTM is applied to automatically achieve fine-grained classification of teachers’classroom questions.The experimental results,i.e.,classification accuracies 95%in 4 MAT questions and 84%in question granularity,shows that this deep learning model can facilitate intelligent sorting of classroom observation data.
作者
黄发良
杨倩
闭应洲
莫运佳
熊冬春
吴兰岸
HUANG Faliang;YANG Qian;BI Yingzhou;MO Yunjia;XIONG Dongchun;WU Lanan(Guangxi Key Lab of Human-Compuer Interaction and Intelligent Decision,Nanning Normal University,Nanning 530100,China)
出处
《福建师范大学学报（自然科学版）》
CAS
2022年第5期43-50,共8页
Journal of Fujian Normal University：Natural Science Edition
基金
国家自然科学基金资助项目(61962038)
广西“十三五”规划教改重点项目(2020AA029)
广西教育科学“十四五”规划重点课题(2022AA18)
南宁师范大学教学改革项目(2020JGX053、021JGX056)。
关键词
教师提问
智慧教育
深度学习
LSTM
teachers classroom questioning
intelligent education
deep learning
LSTM
