Research Of Micro Blog User's Interest Mining Based On Improved Author Topic Model

Microblog is a social network tool in the information age,which can excavate valuable user information from the mass data of microblog.The traditional method of vector-space transforming text into clustering has not been able to deal with the complicated text data of microblog.So many scholars began to use the theme model to carry on the excavation and the analysis to the microblog content,and has greatly promoted the excavation effect of the microblog content,but the previous research did not have enough attention to the microblog intrinsic characteristic and the microblogging user's use characteristic.In this paper,the characteristics of microblog and user characteristics are further analyzed and studied,by choosing the author-topic model to exploit the topic of microblog user interest,and then using the author-topic model as the basis to integrate the repost characteristics of microblog into the model,the microblog generation model AT_IR to excavate the user interest topic of microblog.Through the data collection of the microblog to verify the experiment,the model in this paper on the accuracy of the topic extraction and the F1 Measure value of the model is higher than at the AT model,it is explained that adding the user behavior characteristics of microblog can improve the accuracy of user interest topic mining to a certain extent.