The Research Of Microblog Information Diffusion Model Based On Polarity

Research on information dissemination mechanism of microblog network has important significance, in terms of marketing and public opinion control. Most of the existing information diffusion model focused on the study of text content and user network. But more and more phenomenon indicates that polarity contained by social media content has a positive correlation with the transmission of information. This article’s research purpose is studying the polarity’s role in information diffusion, to improve a new model, and use it to solve the influence maximization problem.Polarity is positive or negative attitude contained in given critical text, polarity analysis is the process of calculating polarity. According to microblog’s writing characteristics and expression form, we put forward a kind of rules combined with supervised learning method to classify microblog and calculate its polarity strength. Through polarity analysis, drawing transmission relationship diagram and regression model, we verified that microblog’s spread popularity is related to the strength of polarity qualitatively and quantificationally. Because of the above conclusion, we add the characteristics polarity into a user behavior prediction based logistic regression model, and propose a new model named polarity combined retweet prediction based influence threshold information diffusion model. Finally, we apply this model to solve the influence maximization problem, and present a topic and polarity based influence maximization estimation method.The experiment results show that the stronger polarity a microblog has, the wider and faster it spread. Polarity combined retweet prediction based influence threshold information diffusion model can achieve good result on simulating microblog dissemination mechanism. Microblog influence maximization method based on topic and polarity also reduces the time complexity and expands the scope of affected node.