Research On The Methods Of Community Vertical Classification Based On Interest

With the rapid development of internet,more and more social media are springing up.Interaction among people becomes more frequent.Community comes into being with its unique structural characteristics in complex social networks.Currently,there are many community classification algorithms in social networks and the algorithm of vertical classification is less.Vertical classification of communities can not only help us learn the structure characteristics of social network directly and solve the problem of locating user requirements in the community,but also offer significant benefits in economic development and marketing in the future.The purpose of this study is to find an effective method for vertical classification of communities in social networks.This thesis regard user interest as a research object is a main method of vertical classification.First,this thesis introduced the definition of interest,community,vertical classification and vertical community.It pointed out that users in the community were divided into two categories through the analysis of the community structure characteristics and community categories.One of them was community leader,the other was ordinary user.Then it analyzed the interest mode and determines the two key issues to deal with for the user interest research in the community.And through the analysis of the limitations of traditional classification algorithm,it determined which direction to use in this research.Next,according to the common interest model representation,it proposes that the representation of interest model can be described from two aspects:qualitative and quantitative.Then,a hybrid interest model was built based on the characteristics of online social networks.In order to verify the correctness of interest model,it validated the model on the basis of a real online social network.In the end,this thesis used social network graph as abstract expression of community structure.It calculated the interest similarity among user nodes by analyzing the user interest.It improved a lot on the basis of traditional community classification standards and proposes a modular calculation method based on interest degree weighting.Compared with the traditional algorithm,this calculation method shows its rationality and advantages from three aspects:the number of community divisions,modularity and interest cohesion.The research base on user interest model that analysis of dynamic characteristics,it can not only dig up user value in social networks,but also provide more accurate advertising and network marketing in ecommerce and play a reference role for the personalized service for users.