User Feature Mining Of Network Community Based On Chameleon Algorithm

The Internet has continuously hit the traditional business model and social cooperation mode.At the same time,the new connection mode of the network community is also exerting a subtle influence on the development of new economy and the progress of society in China.The user is the most important part of the network community,The user is not only the acquirer of web information but also the disseminator and manufacturing source of the website information.The analysis of the massive user data in the network community can be more accurate and faster to grasp the user's some important features,which provide the information foundation for the development of enterprises and companies.Therefore,combining Chameleon clustering algorithm with dynamic modeling,we can find the characteristics of high-quality clusters with different shapes,sizes and densities.It is a worthwhile research to mine the characteristics of users in the network community.In this paper,Chameleon algorithm which does not need human intervention is put forward in the light of the shortcomings of Chameleon algorithm.The improved Chameleon clustering algorithm is proposed by introducing recursive bisection,flood fill and first-jump truncation method.A method that can automatically select the optimal clustering results from the modified chameleon dendrogram is proposed.Then the experiment is validated on data sets of different scales and different types,and compared with three existing improved algorithms,the experimental results show that the improved Chameleon algorithm has a good performance in NMI evaluation criteria,clustering accuracy and running time.Based on the improved Chameleon algorithm,the basic data of the network community are extracted and collected from the user content data and the user behavior data to realize the automatic clustering of the network community and to analyze and explain the clustering results.Then,based on the static and dynamic analysis of the collected user information data source,using WordArt visual tools for user portrait data modeling,in order to describe the characteristics of users,and ultimately based on the constructed user behavior model,respectively,for the product category community and social community put forward precise operation strategy,and provide new ideas and thoughts for solving the operation problem of the current network community.