Research And Application Of Mail Network's Community Detection Based On MCL

At present,people are in the era of Internet media with rapid information dissemination.E-mail has become an important means of communication tool in daily life.Using email to communicate with others,to a certain extent,could reflect real social relationships.The mail network is formed through the communication relationship between people,which contains a large amount of information about the social relationship of the mail users.The mail communication network can be mapped to the epitome of social network relationships.In recent years,in order to more intelligently extract,analyze and process the massive data information as well as unearth potentially useful information in the mail network,the social network analysis technology that is applied to the mail network for community detection has become a hot topic in a new round of research.This paper takes the user communication relationship in the mail network as the research object,and studies the problem of how to accurately identifying the ownership of the nodes in the community.The main contents are as follows:(1)Based on the study of the nature and division of the mail network,this paper proposes a solution for community detection on the mail network based on the undirected graph composed of the mail network.The method uses the Markov Clustering algorithm to analyze the communication behavior between nodes in the mail network.According to the probability change between nodes,the iterative calculation method is used to divide the community.(2)For the traditional Markov Clustering algorithm has the problem of over-fitting to generate small clusters in the process of calculation,which cause the edge nodes that should exist in the community are separated from the whole community and affects the clustering quality,reduced accuracy of community division.This paper proposes a Markov clustering algorithm based on hierarchical agglomerative classes,and get the A-MCL,which uses the cohesion idea in the hierarchical clustering algorithm,combines the Hausdorff formula to merge the existing small clusters,and conducts experiments on the real dataset.By comparing classic community detection algorithm,the proposed algorithm is proved to be effective and can effectively divide the data into communities.These procedures apply the Markov clustering algorithm based on hierarchical agglomerative class to the mail network,sort the Enron dataset,and creat a mail network adjacency matrix.Through experimental verification and analysis,the algorithm can accurately identify the community node attribution and improve the purity of community partitioning.(3)Combined with the Markov clustering algorithm based on hierarchical agglomerative class proposed in this paper,community detection is carried out in the mail network and prototype system of a mail network's community detection is designed and developed.The system is mainly divided into such as data input,data preprocessing and mail network community detection,which further improve the practicability of the algorithm proposed in this paper.