Design And Implementation Of E-mail Network Analysis System Based On Community Detection Algorithm

There exist widely the relation systems either in nature world or the human society.These complex systems are reflected in networks. Being in a large scale, and verycomplex is the relationship of each node, these networks are called COMPLEXNETWORK. As the research of complex network gets deeply these days, a lot offeatures of complex network are discovered by people, of which the most attractive oneis community structure.As the modern science and technology develop rapidly, there comes a closerrelationship among people, and also a more complex relationship network in our lives.They are not only reflected in the expansion of the network’s scale, but also in the morecomplex relation of people. And there are a lot of valuable information in the large-scaleand complex information. By analyzing the networks data, we could obtain a lot ofimportant referential information for our daily life, work, commercial and nationalsecurity. In the same way, as the complexity of the networks of people grows rapidly,the efficiency and accuracy of many previous algorithms becomes less adaptable. Thus,the high-efficient community detection algorithm comes to a research hotspot.According to the description above, we come up with two efficient algorithmsbased on the analysis of the advantages and disadvantages of the existing algorithms.The first one, which absorbed the ideas of layer-dividing of the existing algorithms,moves each node among societies, and then transfers the steady societies intotranscendental points. As applies the Judging-Cut algorithm when moving the nodes, itcuts a lot repetitive computation, so that it shares in a higher time efficiency. The secondone is Overlapping Community Division Algorithm. It detects the core of thecommunity by community Core Detecting Algorithm. In order to obtain the originalstructure of the community, it expands the core of the community which have beendetected by taking advantage of the existing Classical Local Community DetectionAlgorithm. And then, it takes the Amending Algorithm to amend the original community,and obtains the accurate structure of the community. After realizing the algorithms, werealize the Email Network Analysis System using these algorithms. The system administrates the class objects by Spring, and operates the Dada Base by the HibernateStructure, and realizes the conjunction between the algorithms of Java and C++by JNITechnology. It also takes the excellent Prefuse structure to do the GraphingRepresentative works, developing the intuition of the network structure. Meanwhile, inorder to be convenient for users to obtain the statistical information, this system offersthe report forms function, which could save the results into report forms that can usedfor users to look up the details of the results.