基于BiLSTM算法的课程评论情感分类及其成因探究

Emotion Classification of Course Reviews Based on BiLSTM Algorithm and its Underlying Causes

设计了一种基于BiLSTM算法的中文评论情感分类模型,针对在线教育平台的学生评论进行情感分析,分为“Positive”“Negative”和“Neutral”三类。利用网易云课程平台的评论数据,结合Bert预训练模型进行词向量训练。结合SVM分类原理与重采样技术进行模型优化,实验结果显示优化后的模型在精确率、准确率、召回率和F1值上表现优异。通过层次聚类与语义网络分析,可视化展示评论的情感成因,为课程改进提供科学依据。

This study examines the sentiment of students’comments on an online education platform and categorizes them into three categories:"Positive,""Negative,"and"Neutral."It does this by building a sentiment classification model of Chinese comments based on the BiLSTM algorithm.Word vectors combined with the Bert pre-training model were trained using the review data of the NetEase cloud course platform.The model is optimized by combining resampling technology with the SVM classification principle.The optimized model performs exceptionally well in terms of precision,accuracy,recall rate,and F1 value,according to the experimental data.The emotional origins of remarks are represented using hierarchical clustering and semantic network analysis,offering a rationale for curriculum development based on science.