基于全连接神经网络的在线学习行为分类判别

Online learning behavior classification and discrimination based on fully connected neural network

针对教育者难以对学习者多样化的学习行为进行监测和预警问题,提出一种基于全连接神经网络的网络学习者学习行为分类模型,来探究不同学习者的学习特点。首先提取智慧树平台收集的学习者活动数据,剔除当中个人信息部分,选取在线学习行为数据形成数据集;然后进一步清洗数据并对学习行为相关数据进行数据标准化处理;最后搭建全连接神经网络进行学习行为判别。实验结果表明,该模型对于智慧树不同课程中的学习者学习行为分类准确率保持在95.6%以上,与其他神经网络模型相比,该方法在准确率和耗时上均有显著提升,具有很好的应用价值。

It is difficult for educators to monitor and warn the diverse learning behaviors of learners,so a network learner learning behavior classification model based on fully connected neural network(FCNN)is proposed to explore the learning characteristics of different learners.The learner activity data collected by the platform Zhihuishu is extracted,from which the personal information is eliminated,and the online learning behavior data is selected to form a data set.Then the data is further cleaned and the data related to learning behavior is subjected to standardization processing.Finally,an FCNN is built for learning behavioral discrimination.The experimental results show that the classification accuracy rate of the model for learners′learning behavior in different courses of the platform Zhihuishu remains above 95.6%.In comparison with the other neural network models,the accuracy rate of the proposed method has improved significantly and its time consumption is shortened,so it has a certain application value.