Research Of Community Mining On Email Networks

Email is the early form and important component of the social networks, and is apowerful tool for social information exchange and communication. In particular, it playsan important role in the inner information transfer of various types of organizations. Theemail network is one of the important components of the social networks, it could findvarious community structures from chaotic data and reveal relationships between thenodes, and then help people understand the network features better, enhance networkmanagement and information delivery.Through the analysis of structures and propagation characteristics of the email aswell as email system, the paper pointed out the directional characteristics of email networkcommunity, turned directed graph into undirected graph. By drawing on cohesionalgorithm of community division thinking, it improved clustering algorithms of maximalclique, and proposed a community partitioning algorithm based on maximal clique. Firstly,the algorithm finded out all the maximal clique from the email networks; then it mined thecore nodes and built the initial community from the maximal clique, gradually divided thenon-core nodes to the initial community through a certain distance index and similaritydetermination method; eventually it got community division results. In addition, Iachieved the algorithm of parallel computing.I built an email analysis system based on this algorithm, treated the campus internalemail from two aspects of data analysis and community division. The experimental resultsshowed that the algorithm could effectively discover communities, and give a intuitivedisplay of the email community division, thus play a better supporting role in emailmanagement.