基于深度学习的教育政策用户评论细粒度情感分析研究

Fine-Grained Sentiment Analysis of User Comments on Educational Policies Based on Deep Learning

智媒时代微博、抖音等网络社交媒体平台成为政府与公众之间传递信息的重要渠道之一,公众在平台上对教育政策的评论影响着教育政策的实施进程、效果及后续政策的出台。融合主题模型LDA和深度学习模型LSTM,以“双减”政策为例,挖掘面向教育政策的网络社交媒体用户评论,并对其进行细粒度情感分析,剖析用户对教育政策的多维主观情感,为提升教育政策实施效果提供参考。研究发现,网络社交媒体用户对“双减”政策的舆论焦点主要集中在四个主题下的16个评论对象上,其中在素质教育、艺术活动、学历3个方面用户情感偏向于正向;在校外培训、课后服务、教育公平、贫富差距、就业等其余13个方面用户情感偏向于负向。

In the era of smart media,online social media platforms such as Weibo and Tiktok have become one of the most important channels to transmit information between the government and the public,and the public’s comments on education policies on these platforms influence the implementation process,effect and subsequent policies.By integrating the LDA model and the LSTM model,and taking the“double reduction”policy as an example,the study mines the users’comments on education policies on online social medias and finegrainedly analyzes the users’multidimensional subjective emotions towards education policies,so as to provide a reference for improving the implementation effect of education policies.It is found that the focus of online social media users’opinions on the“double reduction”policy is mainly concentrated on 16 comment objects under four themes,among which the users’emotions are positive in three aspects,including quality education,art activities,and academic qualifications;and the remaining 13 aspects are negative,such as out-of-school training,afterschool service,education fairness,rich-poor gap,and employment.