Information Mining On Individuals And Groups In Social Network Of Microblog

As the Web2.0has begun, the social media gives the way of our life anunprecedented level of change. Especially the emergence of microblogrevolutionized the way people get information and made people not only audiencebut also publishers and communicators at the same time. As typical of the socialmedia, microblog allows users customize their own following lists based on interests.Each user can follow others and also be followed, which makes a huge socialnetwork that gives a social property and contributes to the communication ofmessages.In this paper, I will start the microblog social network research mainly from therelationship between uses and group characteristics presented by the social network.I will search for the potential phenomenon of socialization through relation analysisand group mining aiming at promoting the research of social search, informationrecommendation, social e-commerce,microblog marketing, targeted advertising andso on. The main research work of this dissertation can be summarized as thefollowing three aspects.First, I read, parsed and analyzed data of microblog by calling open APIsprovided by Weibo and serialized the data into XML.Second, I built a model for social information and interaction informationbetween Weibo users and demonstrated the correctness of the similarity calculationsof social information by computing the similarity of social information betweenWeibo users, with the interaction information as the standard measure, which fansinformation similarity to that best reflects the intimacy between the users. With thiscomputing method, my experiment with friends recommendation worked very well,which the fans information similarity worked best.Third, based on the similarity of social information, a similarity network wascreated with setting a similarity threshold value. A classic CNM (Clauset、Newman、Moore) group mining algorithm based on the graph cut was employed on thisnetwork and the results was visualized by our self-developed visualization systemof social network analysis. The experiment took NLP users for example, formed theNLP teacher group, NLP enterprise group, NLP student group and several othergroups.