An Evaluation Method Of Microblog User’s Reliability Based On Logistic Regression Model

The research on the reliability of microblog users has become one of the hot points in microblog with the purpose of identifying fake users, giving reasonably and objectively evaluation on microblog users’ identity category. Existing researches could identify ordinary “zombie fans”, but could not identify the new intelligent fake users and rationally evaluate the identity category of microblog users.Logistic regression is one of statistics analysis methods which can be used to predict the identity category of microblog users. For the shortages of the existing methods, we firstly analyze the behavior characteristics of fake users, then reasonably and logically combines users inherent characteristics by the aspects of on-line duration, posting time, posting motivation, posts sources and interaction with others. Finally, a model for evaluating microblog users credibility(WUREM) is proposed based on logistic regression.Experiments adopt Sina microblog as research platform to testify the effectiveness and rationality of our new model. Experimental results indicate that the new model can make an objective and reasonable evaluation and classification on users identity, which effectively solve the limitation and irrationality of traditional methods. The new model not only can recognize the common fake users “zombie fans”, but also has a high recognition rate on new intelligent fake users. Moreover, our research can also provide a reference on the researches of microblog users’ influence, microblog credibility, microblog fever and so on.