基于在线评论的课程思政学生关注点研究
Research on the Concerns of Students in Curriculums with Ideological and Political Content Based on Online Reviews

受众的关注点是教学设计与教学效果评估中的关键问题。教育研究领域关于思政教学中学生关注点的研究较少，且基于问卷的研究方式受到成本、数据量与样本偏差的限制，效果有限。文章提出一种基于深度预训练模型的课程思政评论话题识别方法。基于自主爬取的相关评论文本，首先通过Bert算法进行特征表示和降维；然后针对经典的分层隐树分析方法处理词共现问题的区分度不足及高时间开销等问题，提出了基于阈值的词共现预修剪的分层隐树模型；最后基于提出的方法，实现了数据获取、预处理、话题挖掘及结果展示的全流程平台。实验结果表明，提出算法具有更好的话题检测性能。研究发现，学生对于带有思政内容的课程的关注点主要仍在于课程教学本身，而不满主要对于思政点融入比较机械等情况。
The focus of the audience is a key issue in teaching design and evaluation of teaching effectiveness. There is a lack of research in the field of educational research on the concerns of students in ideological and political education, and the questionnaire based research method is limited by cost, data volume, and sample bias, resulting in limited effectiveness. The article proposes a topic recognition method for course ideological and political comments based on a deep pre training model. Based on self crawling of relevant comment texts, first feature representation and dimensionality reduction are performed using the Bert algorithm; then, in view of the lack of differentiation and high time cost of the classical hierarchical hidden tree analysis method to deal with the problem of word cooccurrence, a hierarchical hidden tree model based on threshold pre pruning of word cooccurrence is proposed; finally, based on the proposed method, a full process platform for data acquisition, preprocessing, topic mining, and result display was implemented. The experimental results show that the proposed algorithm has better topic detection performance. Research has found that students’ focus on courses with ideological and political content mainly lies in the teaching itself, while their dissatisfaction mainly lies in the rigid integration of ideological and political points.