软件导刊2025，Vol.24Issue(9)：34-40,7.DOI:10.11907/rjdk.241486
基于CCSS十年追踪调查的学生留言文本分析
Text Analysis of Student Message Based on CCSS Ten-year Follow-up Survey
摘要
Abstract
Regularly collecting opinions from college students through survey questionnaires is an important measure for universities to im-prove the quality of education.The comments in the questionnaire can largely reflect students' demands,expectations,and evaluations.How-ever,student messages are often fragmented,with issues such as sparse vocabulary and lack of semantic features.To this end,an improved Mini-Batch K-Means clustering algorithm is proposed,which transforms short texts into long texts according to their similarity using cluster-ing methods as inputs for the LDA model.The improved algorithm only uses partial data in each iteration,processes large-scale datasets through random sampling,finds new cluster centers within the sampling batch,and then updates the cluster centers to the global cluster cen-ters,which helps to stabilize the update of cluster centers and provide more accurate clustering results.The experimental results show that the improved clustering method can maintain a running time of less than 10 seconds,and the average PMI value of the model has increased by up to 15.1%.The text analysis results show that in the past 10 years,college students have shown a relatively stable level of attention in terms of employment opportunities,majors,social practice,courses,educational resources,etc.,with individual years highlighting personalized is-sues such as dormitories.关键词
学生留言/Mini-Batch K-Means聚类算法/主题模型/LDA/TF-IDFKey words
student message/Mini-Batch K-Means clustering algorithm/thematic model/LDA/TF-IDF分类
信息技术与安全科学引用本文复制引用
刘哲,单剑锋,杨立军..基于CCSS十年追踪调查的学生留言文本分析[J].软件导刊,2025,24(9):34-40,7.基金项目
国家社会科学基金一般项目(22BTJ030) （22BTJ030）
