基于BPNN的在线学习者元认知能力评估  

Assessment of Online Learners’Metacognitive Ability Based on BPNN

作　　者：王洪江[1] 陈沛瑜 李作锟 张一夫 李南希[1] WANG Hong-Jiang;CHEN Pei-Yu;LI Zuo-Kun;ZHANG Yi-Fu;LI Nan-Xi(School of Information Technology in Education,South China Normal University,Guangzhou,Guangdong,China 510631;Faculty of Education,University of Macao,Macao,China 999078)

机构地区：[1]华南师范大学教育信息技术学院,广东广州510631 [2]澳门大学教育学院,中国澳门999078

出　　处：《现代教育技术》2024年第11期132-142,共11页Modern Educational Technology

基　　金：2023年度广东省教育科学规划课题(高等教育专项)“协作学习中搭便车行为的内部机理分析与干预研究”(项目编号:2023GXJK270)的阶段性成果。

摘　　要：元认知能力被认为是提升学习者在线学习效果的关键因素,其有效评估对于教育教学具有重要意义。但是,当前基于在线学习行为的元认知能力评估方法存在指标选取有差异、不全面且权重模糊等问题。对此,文章首先设计了基于BPNN的学习者元认知能力评估研究框架,通过数据采集和标准化处理,得到19个在线学习行为指标。接着,文章进行了教学实验,通过相关性分析得到11个与元认知能力显著正相关的行为指标,并构建了基于BPNN的元认知能力评估模型。随后,文章进行了模型的性能验证,发现BPNN模型的整体性能和分类性能均最优;同时,文章从行为指标的重要性、影响情况和个体贡献度三个角度进行了模型的可解释性分析,以利于教师做出更科学、合理的教育决策。最后,文章从学习空间、学习资源、教师角色三个方面对BPNN模型的教学应用进行了展望。文章构建的BPNN模型能够准确评估和可视化解释元认知能力,有助于个性化教育教学的开展,从而为教育评估研究提供发展方向。Metacognitive ability is considered to be the crucial factor to improve learners’online learning effect,and its effective assessment is of great significance to education and teaching.However,the current metacognitive ability assessment methods based on online learning behaviors have some problems,such as discrepant index selection,incomplete and fuzzy weight.Accordingly,this paper firstly designed a research framework of learners’metacognitive ability assessment based on back propagation neural network(BPNN),and 19 online learning behavioral indicators were obtained through data collection and standardized processing.After that,a teaching experiment was conducted and 11 behavioral indicators that significantly positively correlated with metacognitive ability were obtained through correlation analysis,and the assessment model of metacognitive ability based on BPNN was conducted.Subsequently,the performance of this model was verified.It was found that the overall performance and classification performance of BPNN model were the best.At the same time,the interpretability of the model was analyzed in this paper from three perspectives of the importance,influence situation and individual contribution of behavioral indicators,so as to help teachers make more scientific and reasonable educational decisions.Finally,the pedagogical application of the BPNN model was looked forward from three aspects of learning spaces,learning resources,and teachers’role.The BPNN model constructed in this paper can accurately evaluate and visually interpret metacognitive ability,which was helpful to the development of personalized education and teaching,and thus provides the development direction for educational evaluation research.

关 键 词：元认知能力 BPNN 行为指标 算法模型 可解释性 

分 类 号：G40-057[文化科学—教育学原理]