Affective Computing And Public Opinion Analysis Of Micro-blog Topics

Nowadays,people are more accustomed to keeping up with current affairs and expressing their views through the social network.With the growing popularization of microblog,more and more users begin to pay close attention to and use micro-blog,micro-blog become the most popular network platform at present.The microblog topic provides an excellent platform for people to discuss popular events.Microblog subject can provide a large number of effective data for the current emotional analysis and public opinion analysis research,so the study on microblog topic is a research hotspot at present.This article mainly uses the microblogging topic data to realize the analysis of people's emotion and the analysis of public opinion.The traditional method of sentiment analysis is based on affective lexicon,rules and machine learning to complete.In this thesis,the emotional dictionary and machine learning are combined to complete the sentiment analysis of micro-blog topic.And compared with the traditional micro-blogging emotional analysis which only considers micro-blogging text content,this thesis take into account the emoticons,and will do research on micro-blogging topic of emotional analysis from two aspects—— the micro-blogging text emotional analysis and micro-blogging emoticons emotional analysis.Firstly,choose the Chinese emotional words ontology library as the text emotional thesaurus,using naive Bayesian to complete the micro-blogging text part of the emotional tendencies to calculate.And then,establish the emoticons library through expression symbol clustering algorithm implemented based on the FP growth algorithm and retrieve the distance emoticons clustering algorithm,and the emotional tendencies of the part of the emoticons in the micro-blogs are also calculated by the naive Bayesian algorithm.At last,combine the two aspects and we will get the emotional tendencies of micro-blog topics.Experiments show that the emotional tendencies of micro-blogging topics,which are taken into account in this thesis,are more accurate for micro-blogging emotional tendencies.This article about evolution analysis of micro-blogging topic public opinion is to combine the calculation result of micro-blogging emotional tendencies and the diffusance of micro-blogging topic to get public opinion value of micro-blogging topic,and then further complete the micro-blogging topic public opinion evolution analysis.The diffusance of the micro-blogging topic is calculated based on the micro-blogging topic forwarding,point of praise and the amount of comments,which micro-blogging forwarding is based on the logic regression model of micro-blogging forward to predict,the number of users whose results are calculated by forwarding the prediction model is higher than 50% as the forwarding amount of the micro-blogging topic.Micro-blog topic public opinion evolution analysis takes into account the time,and divide the micro-blogging topic into time slices,and by improving the traditional theme model LDA model to establish micro-blogging topic model MTLDA,according to MTLDA to calculate each time film theme.Finally,based on the KL distance to calculate the relationship between the adjacent time slices,and analyze the evolution of micro-blogging topic public opinion over time.Through the experiment of multiple topics,it is proved that the method of public opinion analysis is more accurate and more timely.