Sentiment Analysis And Propagation Rules Of Micro-Blog Hot Topics

As an open social media,micro-blog has been accepted by more and more users,this platform is a grassroots platform,users can be arbitrary views,share information,so its role in public opinion spread there for all to see.Micro-blog hot topic in the real-time delivery,point forward,reviews,and user praise without restriction,make the news propagation speed very fast,in this case,user's opinions in the micro-blog content is preserved in the form of long-term,personal emotions can be multiplied,if there is not timely guidance,it is easy to form a negative opinion,even cause the malignant event in real life.Therefore,micro-blog managers need to monitor the hot topics in real time and guide public opinion in the right direction before the occurrence of group polarity events.Based on the above background,this paper will study propagation law of two types of the SINA,micro-blog and entertainment hot topic,compared to two topics in the aspect of heat grow along with the time difference,lastly,a variety of methods to analyze topic tweets emotional polarity.The main research work includes:first,the extraction of effective features of SINA hot topics and the establishment of topic heat formula;Second,the drawing and fitting of the heat curve of SINA hot topics and the mining of topic propagation rules.This study was carried out on the SINA web crawler,crawl and various features extracted in the EXCEL file,according to the formula to calculate heat value,using SPSS software to topic heat value to simulate curve drawing and change over time.By comparing the similarities and differences between the growth curves of two kinds of hot topics,the propagation rule of two kinds of topics is explored.Third,SINA hot topic blog post emotional polarity discrimination.There are four algorithms used in this study:(1)a dictionary based micro-blog emotional polarity analysis;(2)analysis of emotional polarity based on naive bayes;(3)analysis of emotional polarity based on TextRank and dictionary;(4)analysis of the emotional polarity of microblogs based on TextRank and naive bayes.The innovation of this study point has two:first,through the heat topic trend research in the limited nature,put forward the method that the starting point of the topic as the start,the number of hours of the topic as the variable,the cumulative heat as the dependent variable.Second,proposed text sentiment analysis method based on TextRank and dictionary and based on TextRank and naive bayes,and compared with the traditional sentiment analysis method,analyzed its advantages in accuracy and efficiency.The purpose of this study is to predict the trend of hot topics and to determine whether a topic will have a serious negative impact,so as to form a rapid and effective method of public opinion supervision.