Topic Behavioral Study of Microblog Content

With the exponential growth of user generated messages, twitter has become a social site where millions of user can exchange their opinion. It is very difficult to find the viral topic from the twitter. So in order to find out this, we adopt both behavioral factors and sentimental factors like positivity and negativity of a content. Sentimental analysis aims to aggregate and extract emotions and feelings from different types of documents. We develop a framework for analysing and modelling contents using behavioral factors such as topic virality, user virality, user susceptibility and sentimental analysis. Content is undergone through sentimental analysis process using Stanford Natural Language Processing tool (NLP). We develop a factorisation method to simultaneously derive the three sets of behavioral factors and use latent Dirichlet allocation algorithm for topic generation. We use parametric learning to mine the behavioral factors and sentimental factors and predict the virality of blog content. Experimental result shows that our method can effectively find the content virality in a microblogging site.