Research On Micro-blog Interest Community Detection Method Based On Content Similarity

With the popularity of the network and the rapid development of communication technology,the public access to the Internet is more convenient.More and more people get a lot of information through the network and complete the efficient dissemination of information.This social situation has promoted the rapid development of micro-blog.According to the forty-first China Internet development statistics report,by the end of September 2017,the number of active users in China's micro-blog was 316 million.In micro-blog,such a huge user community has generated huge user data.The value of micro-blog user data is not a disorderly thing,but needs to spread the data analysis object to a community as a unit,need to find and excavate the community,and then obtain valuable information in the micro-blog user data.Community discovery in a complex micro-blog network has important theoretical and practical values for the improvement of micro-blog's personalized recommendation systems precision advertising,and enterprise marketing.So the content of this paper is how to accurately understand the user's interest,detect the user's interest community,and quickly and effectively detect the users with similar interests.This paper has completed the following aspects:(1)On the premise of analyzing and introducing the research background and significance of micro-blog interest community discovery,this paper expounds the network representation,network properties of micro-blog,and the structure of microblog network.At the same time,it summarizes and expounds the current research status of text representation model and text similarity calculation method,and introduces AP clustering at the same time.The algorithm principle and implementation steps of algorithm and Newman fast algorithm are discussed.(2)By analyzing the content characteristics of micro-blog users,this paper constructs interest representation methods for interest leaders and ordinary users respectively.In the study of interest leader recognition,this paper puts forward a new user influence index system through the content of micro-blog users' behavior,microblog text content and user social content..By analyzing the concern of users,this paper proposes the use of Page Rank algorithm to modify the influence of users,and finally realizes the knowledge of the user's opinion leaders.(3)In this paper,a content based micro-blog interest community discovery framework is constructed.On the basis of the identification of user interest leaders,the article uses AP algorithm to realize the discovery of the core user interest community,and then through the fusion of Newman fast algorithm idea,the final micro-blog user community recognition is completed by judging the change of the module degree value of the other common user nodes joining the core user interest community.(4)Finally,by comparing the method of community discovery in this paper with the GN algorithm and the LPA algorithm,it is proved that the method proposed in this paper can well complete the discovery of micro-blog interest community.In the contrast experiment,we introduced the bad environment of the experiment,the way of collecting micro-blog data and the evaluation index used in the experiment.