Sentiment Analysis Of Danmaku On Bilibili Website From The View Of Interactive Ritual Chain Theory

Danmaku,as a new form of online commentary,enriches the way users communicate with each other.The real-time nature of Danmaku enables real-time interaction between users,reflecting their real-time emotions and evaluations of the video.The Interaction Ritual Chain theory is a theory that studies situations and processes.Based on this theory,analyzing the emotions in Danmaku can not only enrich the research on emotion analysis and improve natural language processing technology,but also explore the commercial value of Danmaku interaction ritual chains,and provide ideas for the commercial development of the platform.This article starts with the Interaction Ritual Chain theory proposed by Collins,and based on the elements of the interaction ritual chain and the results it brings,combined with the characteristics and usage scenarios of Danmaku on Bilibili,adds offline behaviors as a result to the interaction ritual process of Bilibili,thus constructing the Bilibili interaction ritual chain model,laying the foundation for subsequent emotion analysis.In order to study the emotional changes of variables related to the Bilibili interaction ritual chain,this study first uses the TF-IDF algorithm to obtain keywords to identify the focus of user attention;secondly,uses the BTM model to cluster users’ emotions to discover shared emotional states;then constructs a sentiment dictionary to analyze the proportion of emotions in Danmaku,selecting the Dalian University of Technology’s Chinese emotion dictionary ontology library as the basic dictionary,introducing a multi-dimensional emotion dictionary for Danmaku,marking the emotional intensity and polarity of each word,and expanding it into a comprehensive sentiment dictionary.The formula for calculating the emotional value is determined by considering the influence of emotional polarity,emotional intensity,negation words,adverbs of degree,and their relative positions on the emotional value through the change of emotional value.Finally,the Bilibili Danmaku information emotion analysis model is constructed from the perspective of interaction ritual.The research data selected the Danmaku of "He Tongxue",a top uploader in the Bilibili technology zone,for empirical analysis.He Tongxue’s videos were divided into three categories,and the most popular Danmaku of each category was crawled using Python language as data samples.Based on the emotion analysis model,the proportion of emotions in the Danmaku text was analyzed to determine the specific expressions of group cohesion,group symbols,moral sense,and offline behavior in the results of the interaction ritual.The core of the interaction ritual is to stimulate individual emotional energy.This study divided the Danmaku into time intervals,used the Matplotlib library to draw a curve of the emotional value time series,and thus more intuitively depicted the changes in user emotional energy.The study found that users in He Tongxue’s videos had a higher proportion of positive emotional tendencies,such as "good" and "happy",followed by negative emotions such as "surprised" and "sad",indicating that users gained more positive emotional energy.This can help uploaders recognize the emotional tendencies of users,reasonably guide interaction,and increase video traffic.This study also found that Danmaku contains effective emotional information.Clustering emotions in Danmaku can generate user emotion labels and the maximum points in the emotion value time series graph can be used as hotspots in the video,providing ideas for developing new ways of video retrieval and recommendation.