【作 者】：
王保华， 熊 余， 姚 玉， 储 雯， 吕 翊
【关 键 词 】：
学生教学评价； 情感分析； 深度学习； 深度记忆网络； 卷积神经网络
【栏 目】：
课程与教学
【中文摘要】：
随着教育信息化建设的深入推进，教学系统中积累了海量的学生教学评价数据，这些数据蕴含了丰富的信息，亟待挖掘利用。为了挖掘学生教学评价中的情感倾向，为提高教学质量提供科学依据，文章提出了一种基于双通道深度记忆网络的深度学习模型，用于学生教学评价的方面级情感分析。在该模型中，设计了双通道策略以充分提取评语中隐含的局部特征和上下文依赖信息，并使用循环注意力机制提取与特定教学方面相关的情感信息以实现细粒度的方面级情感分析。通过在真实的教学评价数据集上进行实验，结果表明，所提出的方法能有效挖掘学生评价中关于不同教学方面的情感倾向，为教师和教学管理者了解并改进教学提供依据。
【英文摘要】：
With the further development of education informatization construction, a huge amount of student teaching evaluation data has been accumulated in the teaching system, which contains rich information and need to be mined and utilized urgently. In order to find out the affective tendency in student teaching evaluation and provide a scientific basis for improving teaching quality, this paper proposes a deep learning model based on a double-channel deep memory network for sentiment analysis in student teaching evaluation. In this model, a double-channel strategy is designed to fully extract local features and context-dependent information implied in comments, and a circular attention mechanism is used to extract the emotional information related to specific teaching aspects to achieve fine-grained sentiment analysis. The experimental results on real teaching evaluation dataset show that the proposed approach can effectively mine the affective tendency of different teaching aspects in students' evaluations of teaching, providing a basis for teachers and teaching administrators to understand and improve teaching.