Analysis Of Social Networks Based On Stochastic Block Models

With the process of information technology,network data has received a lot of attention,virtual network and real life has become increasingly inseparable relationship.The community structure represents the set of specific objects in the network,reflecting the inherent laws of a particular group,and then exploring the hidden value of these special groups.But the amount of network data is growing,the structure is more and more messy,if only the traditional way to performance,understanding is very difficult,it is difficult to network data in the overall display.This brings the challenge to the network data analysis,but also gave birth to a new statistical method to analyze the network data.This paper firstly analyzes the social circles based on E-mail data and analyzes the characteristics of network graphs.Then,the hierarchical clustering algorithm is used to obtain the initial clustering number,and then the clustering clustering based on k-means is used to obtain the more accurate clustering number.The basic knowledge of spectrum clustering algorithm in clustering is analyzed,and the network data is analyzed by spectral clustering method.Finally,a number of different community structures are obtained.Finally,the stochastic block model algorithm is used to estimate the community block number of the network data,and the clustering under the random block model is obtained by comparing with the algorithms described above,and it is applied and verified in E-mail network data The Finally,the results of the algorithm are summarized and the optimal clustering number is obtained.And the community block of the various algorithms in the future improvements made a vision.