Sentiment Classification Analysis Of Barrage And Comment Based On Deep Learning And Its Application

With the continuous development of the network video industry,more and more new elements join it,which adds vitality for its vigorous development.The barrage is an important part of it.The launch of the barrage mode creatively opens up the routes for users to participate in watching videos,and greatly improves the interaction frequency and activity of users.With the widespread launch of the barrage mode on video websites,its audience is expanding step by step,and the commercial value of the bullet screen itself is also constantly improving.Then,what are the differences between the barrage screen model and the traditional comment model,and how the two affect the spread of online video,which are the problems that this paper wants to study.This article first crawled the text information of about 1.08 million barrages and comments from 82 videos on Bilibili website(hereinafter referred to as Station B).Then feature extraction is carried out based on Word2 Vec to help the subsequent neural network better understand the semantics.Secondly,the LSTM neural network is used to build a classifier for sentiment classification of bullet screen and comments.According to the model effect,one-way classifier and two-way classifier are selected for sentiment classification of barrages and comments.Based on the extension of the above results and the existing communication theory,this paper extracts the indicators in the network video communication path,and makes an empirical analysis of the possible influencing factors.Finally,a network video communication model with "cognition-emotion-behavior" as the first level classification is established.It helps us to explore the influence of different user cognition and different emotional tendencies on network video communication and the differences among various factors.The results show that barrage and comments show different emotional tendencies.The barrage focus on meeting the interaction and participation needs of users,while the comments focus on meeting the emotional expression needs of users.The "cognitive" level represented by the "number of bullet screens","number of collections" and "number of coins" has a higher influence on the communication of online video than the "emotion".The way of impact of barrage and comments on video transmission is similar,but there are differences in effect.The negative affective tendency of both can improve the communication effect of video,while the positive affective tendency is opposite.The effect of comment emotional tendency on video communication is higher than that of barrage.