基于情感词典和机器学习的微博情感分析

北京理工大学珠海学院2020届本科生毕业论文Micro blog emotion analysis based on emotion dictionary andmachine learningAbstractNowadays,the Internet age has been fully integrated into our life,and we can under-stand the world at home,and micro blog is one of the main ways to spread information.In2019,the proposal of"996"has aroused hot discussion among Chinese netizens,and microblog users have also put forward their views.In order to know whether microblog userslike or dislike "996",I use the deep learning method to build an emotional analysis modelof microblog comments and analyze the emotional preferences of user comments.I have established three deep learning models:LSTM model,BILSTM model andGRU model as a comparison to find the most appropriate emotion analysis model.Aftercomparing the accuracy,efficiency and structure of the three models,it is found that bilstmmodel is the best for Chinese processing,and the highest accuracy is 82.65 while theefficiency is not much different from the other two models.Therefore,I decided to chooseBILSTM model as Weibo comment emotion classification model.Iimported 10286"996"comments into the BILSTM model for emotional analysis,andobtained 7222 negative ones,151 neutral ones and 2913 positive ones.The results showthat microblog users are basically opposed to the996"view,because the"996"view meansthat workers need to work longer and bring more pressure,so most people are opposed.Forthose who are engaged in network related businesses,they should reasonably arrange resttime and properly reduce pressure."996"is only tentative in the scope of discussion,andimplementation should be cautious.Keywords:LSTM model BILSTM model Emotion analysis GRU model