Identification of micro-blog opinion leaders based on user features and outbreak nodes

Abstract

At present, recognition of micro-blog opinion leaders mainly depends on the number of users posting micro-blogs, registration time, the number of good friends and other static attributes. However, it is very difficult to obtain the ideal recognition results through the above mentioned methods. This paper puts forward a new method that identifies the opinion leaders according to the change of user features and outbreak nodes. Deeply analyzing various attributes and behaviors of users, on the basis of user features and outbreak nodes, user's attribute features are regarded as the input variables, behavior features of the user and outbreak nodes are regarded as observed variables. The probability as an opinion leader is the latent variable between input variables and observation variables, and the constructed probability model is used to recognize micro-blog opinion leaders. Experiments are carried out on the two real-world datasets from Sina micro-blog and Twitter, and the comparative experimental results show that the proposed model can more precisely find the micro-blog opinion leaders.