基于深度学习的体育慕课教学诊断与优化

An approach to instructional diagnosis and optimization of sports MOOCs based on deep learning

摘要

教育数字化转型与全民学习型社会建设深入推进,为体育在线课程的智慧化发展提供了重要契机.运用文本挖掘法、案例分析法等研究方法,从学习者视角切入,对中国大学慕课平台185门体育慕课课程的学习者评论数据进行了深入挖掘,探讨卷积神经网络-循环神经网络混合模型进行情感分析,聚焦负向评论进行教学诊断;构建潜在狄利克雷分布主题分析模型,精准拆解问题剖析致因;提出针对性解决方案并明确各路径的解决成效.研究发现体育慕课核心短板与教学痛点:1)课程设置与学习者需求错位;2)课程分类与检索粗放;3)课程内容静态化与窄化;4)教学与考核机制脱节;5)反馈支持体系不完善.因此,提出基于情感主题分析的教学表达与呈现优化;基于主题与情感标签的课程内容智能适配;融合具身认知的实践教学环节创设;基于多模态数据的个性化学习支持与情感干预.

Abstract

The deepening advancement of educational digital transformation and the construction of a lifelong learning society have provided significant opportunities for the intelligent development of online physical education courses. Employing research methods such as text mining and case analysis, this study adopts a learner-centered perspective, conducting an in-depth mining of learner review data from 185 sports MOOC courses on the Chinese university MOOC platform. It employs a CNN-RNN hybrid model for sentiment analysis, focusing on negative reviews for instructional diagnosis, and constructs an LDA (latent Dirichlet allocation) topic analysis model to accurately deconstruct problems and analyze their causes. Targeted solutions are proposed, along with an evaluation of the effectiveness of each approach. The research identifies shortcomings and teaching challenges in sports MOOCs: 1) misalignment between course design and learner needs; 2) crude course classification and retrieval systems; 3) static and narrow course content; 4) disconnection between teaching and assessment mechanisms; and 5) an inadequate feedback and support system. Accordingly, Optimizing Teaching Expression and Presentation Based on Sentiment-Topic Analysis; Intelligent Adaptation of Course Content Based on Topic and Sentiment Labels; Designing Practical Teaching Segments Integrating Embodied Cognition; Personalized Learning Support and Emotional Intervention Based on Multimodal Data are proposed in the study.

关键词

体育慕课 / 学习者评论 / 文本挖掘 / 教学诊断 / 精准优化 / 卷积神经网络 / 循环神经网络 / 潜在狄利克雷分布

Key words

sports MOOCs / learner reviews / text mining / teaching diagnosis / targeted optimization / CNN / RNN / LDA

参考文献

[1] 教育部 : 慕课超过9.7万门,学习人数达13.9亿人次,我国慕课数量和学习人数均居世界第一[R/OL].(2024-12-16)[2025-03-11]. http://www.moe.gov.cn/jyb_xwfb/s5147/202412/t20241217_1167307.html.

[2] 国务院 印发《关于深入实施“人工智能+”行动的意见》[R/OL]. [2025-08-26]. https://www.gov.cn/yaowen/liebiao/202508/content_7037868.htm.

[3] 王锐, 王锦萱, 罗仙平 . 学习评论视域下基于KANO模型学习实效评价指标体系构建与应用实践[J]. 黑龙江高教研究, 2021, 39(10): 155-160.

[4] 王丽坤, 王宏, 陆玉昌. 文本挖掘及其关键技术与方法[J]. 计算机科学, 2002, 29(12): 12-19.

[5] Wang L K, Wang H, Lu Y C. The text mining and its key technigues and methods[J]. Computer Science, 2002, 29(12): 12-19.

[6] Mihalcea Rada . The text mining handook: Advanced approaches to analyzing unstructured data[J]. Computational Linguistics, 2008, 34(1): 125-127.

[7] Justicia de la Torre C, Sánchez D, Blanco I, et al. Text mining: techniques, applications, and challenges[J]. International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2018, 26(4): 553-582.

[8] Kayser V, Blind K . Extending the knowledge base of foresight: The contribution of text mining[J]. Technological Forecasting and Social Change, 2017, 116: 208-215.

[9] 王婷, 杨文忠. 文本情感分析方法研究综述[J]. 计算机工程与应用, 2021, 57(12): 11-24.

[10] Wang T, Yang W Z. Review of text sentiment analysis methods[J]. Computer Engineering and Applications, 2021, 57(12): 11-24.

[11] Blei D M, Ng A Y, Jordan M I . Latent dirichlet allocation[J]. Journal of Machine Learning Research, 2003, 3: 993-1022.

[12] 綦方中, 田宇阳. 基于BERT和LDA模型的酒店评论文本挖掘[J]. 计算机应用与软件, 2023, 40(7): 71-76.

[13] Qi F Z, Tian Y Y. Hotel review text mining based on bert and lda model[J]. Computer Applications and Software, 2023, 40(7): 71-76.

[14] 张义祥. 基于文本挖掘技术的第十四届全运会网络舆情研究[D]. 西安: 西安体育学院, 2023: 30-35.

[15] Zhang Y X. Research on network public opinion of the 14th national games based on text mining technology[D]. Xi'an: Xi'an Physical Education University, 2023: 30-35.

基金资助

国家社会科学基金(24BTY081)