Dynamic Analysis Of Weibo Topic Emotional Tendency Based On Machine Learning

With the evolution and popularization of the Internet,social networking on the Internet has become an indispensable aspect of people’s lives.Among many internet social products,micro blog has become the largest and most influential platform in China’s society nowadays.Weibo topics can have a large number of emergencies which can influence strongly,spread rapidly and have a long duration.Therefore,dynamic analysis of the public’s emotional tendencies in Weibo topics will help relevant authoritative media and government agencies to monitor and guide public opinion to avoid the occurrence of rumors and vicious transmission incidents.First of all,this paper selects Hong Kong demonstration topics as the research object.Through crawler technology,it finds and gleans relevant blog posts and comment data on this topics.To prepare for the pre-classification of emergencies and sentiment analysis of comments,it cleans the data,removes duplicate words,cuts sentences into words and eliminate irrelevant words.Secondly,this paper constructs a dictionary that pre-classifies emergencies.Based on this dictionary,this paper pre-classifies emergencies through making word cloud pictures and word frequency statistics to categorize emergencies into three types:tending to lead to positive,tending to lead to negative and neutral.Then this paper analyzes the sentiment of the comment data by dictionary-based method and the naive Bayes methods.After comparison,it is found that the results obtained by the naive Bayes sentiment analysis method can have higher accuracy.Finally,this paper uses knowledge graph related technology to construct emotional knowledge graphs of emergencies and Weibo topics.It also conducts a dynamic analysis of the public sentiment tendency on Hong Kong demonstration topics by combining the results of pre-classification of emergencies with the results of sentiment classification.The results found that in different periods of the development of these topics,the guiding influence of the emergencies itself and that of authoritative media on the public are different.The relevant authoritative media and government agencies need to base on characters of different development stages of Weibo topics to explain and lead public opinion at key emergency nodes to prevent the occurrence of rumors and vicious transmission events.