| In recent years,with people’s increasing demand for a better life,entertainment activities have become rich and colorful,and movies have become the seasoning in people’s daily life.The booming development of the Internet has promoted the rapid development of the film industry.Online platforms allow audiences to post their own comments on movies after watching movies.These movie comments contain audiences’ emotional attitudes and willingness to recommend movies,which are significant for analyzing movies and the box office.predictions.By analyzing the diverse movie review data in the network,the attitudes of moviegoers and potential movie-watching users can be mined;The movie elements that a movie is loved by the viewers can be effectively analyzed through the sentiment analysis of movie reviews,contributing to promote the development of the film industry and correctly guide the user’s choice tendency.At present,the sentiment analysis research of movie reviews focuses on the reviews after watching the movie.This thesis conducts text sentiment analysis research on movie barrage reviews,analyzes the data characteristics of movie barrage,and mines the emotional information contained in movie barrage:(1)In the light of the real-time characteristics of bullet screen,this paper designed crawler technology to obtain movie bullet screen comments from video websites,uses data cleaning,word segmentation and data annotation to process data,visualizes the movie bullet screen,highlights the key words in the movie bullet screen,and studies the plot,actors and other factors on the emotional tendencies of movie barrage.(2)In view of the characteristics of movie barrage with sparse sentences and many buzzwords,this paper uses the mutual information algorithm of emotional tendency points to construct an emotional dictionary in the field of movie barrage,and uses the combination of emotion scoring mechanism and manual annotation to analyze the emotional tendencies of movie barrage.(3)Due to the large amount of data and wide-ranging content of movie barrage reviews,this paper proposes a barrage sentiment analysis method based on Attention-Bi LSTM-CNN on the basis of the Bi-LSTM model.In the process of text vector representation,an attention mechanism is added to capture key emotional information.And a CNN module is introduced to further improve the feature extraction capability of model information.The experimental results show that the sentiment analysis method based on Attention-Bi LSTM-CNN has better effect in the sentiment analysis task of movie barrage. |