摘要
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.
作者
徐锐
杨帆
XU Rui;YANG Fan(Computer and Information Engineering College,Hubei Normal University,Huangshi Hubei 435002,China;School of Computer Science and Engineering,Hubei University of Education,Wuhan 430205,China)
出处
《湖北第二师范学院学报》
2024年第8期34-42,共9页
Journal of Hubei University of Education
基金
2019年度湖北第二师范学院校级教研项目(X2019011)。
关键词
BiLSTM算法
中文课程评论
情感分析
层次聚类
BiLSTM algorithm
Chinese course review
sentiment analysis
hierarchical clustering
