摘要
针对目前很多在线学习平台学习行为数据采集不充分,数据特征量不足的问题,通过选取高校成人教育平台学生学习日志数据,建立深度学习成绩预测模型,比较各类深度学习预测模型与传统的统计学习预测模型的有效性,发现全连接深度神经网络模型效果最好。而且发现加入注意力机制,并不能提升预测模型的准确率。在此基础上,探索了以API方式封装模型,并将封装模块嵌入互联网在线学习平台,平台系统将对日志数据进行提取,预测学习结果,进行学习反馈预警,从而改善学生的在线学习方式和学习效果。
For the current many online learning platform learning behavior data collection is not sufficient,the problem of selecting adult education platform students learning log data,establish deep learning performance prediction model,compare all kinds of deep learning prediction model and the effectiveness of the traditional statistical learning prediction model,found that fully connected deep neural network model is best.Moreover,it is found that adding the attention mechanism can not improve the accuracy of the prediction model.On this basis,the paper explores the API packaging model,and the packaging module is embedded in the Internet online learning platform.The platform system will extract the log data,predict the learning results,and conduct learning feedback and early warning,so as to improve students’ online learning methods and learning effect.
作者
薛艳肖
徐润森
曾庆宇
XUE Yan-xiao;XU Run-sen;Zeng Qing-yu(Jiangsu Open University Business School,Nanjing 210036,China;Nanjing University of Posts and Telecommunications School of Management,Nanjing 210003,China)
出处
《电脑与信息技术》
2022年第6期4-6,10,共4页
Computer and Information Technology
基金
2017江苏省高校哲学社会科学研究重点项目(项目编号:2017ZDIXM169)
2021江苏省高校哲学社会科学研究项目(项目编号:2021SJA0766)。
关键词
在线学习
深度学习
成绩预测
机器学习
online learning
deep learning
performance prediction
machine learning
