一种基于特征偏移补偿的深度智能化教学评价方法

A Deep Intelligent Teaching Evaluation Method Based on Compensation for Feature Deviation

针对慕课(MOOC)评论中存在少数类特征偏移的问题,提出一种基于特征偏移补偿的深度智能化教学评价方法.该方法首先使用Glove预训练模型获取MOOC评论的分布式词向量;然后采用浅层卷积神经网络,通过多个卷积核学习教学评价的语义,引入不同类别评论的数量设计影响因子,归一化该影响因子并应用到交叉熵损失函数中;最后基于Coursera平台的本科学生教学评论数据集,通过与其他损失函数在F_(1),gmean, balance, gmeasure等评价指标上进行性能对比实验.实验结果表明,基于归一法的特征偏移补偿损失函数在gmeasure指标上比基类损失函数得到了最多15.40%的性能提升,并且采用该损失函数的分类模型也表现出较强的稳定性.

Aiming at the problems that there were feature deviation for the minority class in massive open online course(MOOC)reviews,we proposed a deep intelligent teaching evaluation method based on compensation for feature deviation.Firstly,this method used the Glove pre-training model to obtain the distributed word vectors of MOOC reviews.Secondly,the shallow convolutional neural networks were used to learn the semantics of teaching evaluation through multiple convolution kernels.The number of different types of reviews was introduced to design influence factors,which was normalized and applied to the cross-entropy loss function.Finally,the data set of undergraduate teaching reviews based on Coursera was compared with other loss functions on F_(1),gmean,balance,gmeasure and other evaluation indicators.The experimental results show that the loss function based on normalized feature deviation compensation has a performance improvement of up to 15.40%on gmeasure than the base loss function,and the classification model using this loss function also shows strong stability.