基于双向LSTM和LDA的新冠肺炎情感分类和热门主题挖掘方法  

COVID-19 Sentiment Classification and Hot Topic Mining Method Based on Bi-LSTM and LDA

作　　者：王学贺[1] 李晓磊 赵华[2] Wang Xuehe;Li Xiaolei;Zhao Hua(Division of Computer Science,Heze Medical College,Heze 274030,China;College of Computer Science and Engineering,Shandong University of Science and Technology,Qingdao 266590,China)

机构地区：[1]菏泽医学专科学校计算机教研室,山东菏泽274000 [2]山东科技大学计算科学与工程学院,山东青岛266590

出　　处：《宁夏大学学报（自然科学版）》2022年第3期304-308,317,共6页Journal of Ningxia University(Natural Science Edition)

基　　金：山东省自然科学基金面上项目(ZR2021MG038)。

摘　　要：面对新冠肺炎疫情带来的重大影响,通过自然语言处理技术,深入挖掘民众关于新冠肺炎的观点与看法,为疫情期间政府应对网络舆情危机提供参考.针对当前研究中将主题和情感孤立研究的缺陷,首先从微博上收集网络舆情的相关数据,然后通过LDA主题模型和基于Bi-LSTM的情感分类方法进行主题-情感的融合分析.结果表明,Bi-LSTM模型可以较好地识别出喜、怒、哀、惧4类情感,同时LDA主题模型在热门主题挖掘方面也表现良好.Due to the severe impact brought by the COVID-19,natural language processing techniques are used to deeply explore and analyze people’s views and concerns,so as to provide reference for the government to respond to online public opinion crisis during the epidemic.To address the shortcomings of current research that usually studies topics and sentiments in isolation,relevant data on online public opinion from Weibo were collected,and then a topic-sentiment fusion analysis through LDA topic model and Bi-LSTM-based sentiment classification method was conducted.The results show that the Bi-LSTM model can better identify the four emotions of happiness,anger,sorrow and fear,and the LDA topic model also performs well in hot topic mining.

关 键 词：新冠肺炎 网络舆情 双向LSTM 主题模型 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]