Key Technologies Research On Community Structure Of Network

With the development of information technology, highly interconnected networkspervade both our society and the information world around us. Microblogs,facebookand QQ capture complex relationships among people, making them importantcomponents of people’s daily life. A social network is an abstract concept consisting ofthe set of people with relationships linking pairs of humans. A more general socialnetwork, namely, natural network, contains the gene network, citations network and soon. These networks contain muc h impor tant information. The study of natural networkshas become a hot research topic. The paper focuses on community analysis and linkanalys is ranking.The most important feature of the natural network is the clustered structure, whereedges between vertices in the same cluster are dense but inter structure edges are sparse.Accurate identification of clustering structure of networks called community detection,which is widely used in protein interaction analysis, electronic commerce, terroristorganization identification and other fields. Firstly, this paper introduces classicalcommunity detectionalgorithms.Based on the clustered-first traversal, as well as a novelevaluation, the paper proposes an improved community detection algorithm with lowercomplexity. Experimental results on benchmark data verify the effectiveness of thealgorithm.Traditional community detection algorithms focus on the entire data of network.However, as the result is the community structure of the entire network, it costsexcessive computational resource to get the result. At the same time, most of thecommunity does not make sense to the user.Moreover, it is always not practical to getthe complete data for a natural network. Therefore,a local community detectionalgorithm with part of the data is proposed,based on the cluster-first traversal andsecondary cut. The algorithm was tested on benchmark and synthetic datasets.Testresults show that the algorithm can identify local community structure of seed verticeseffectively with lower complexity.As vertices of a network are not equally important, it is meaningful to identify theimportance of the vertices in a network. Link analysis ranking can be widely used in thesearch engines, literature impact factor, identify the important member of terroristorganization and other fields. After a background introduction link analysis ranking, thispaper analyzes the classical algorithm for bridge vertices rank. Then, a random walkcentrality algorithm which has excellent performance was analyzed. Focusing on itsmain complexity, a fast algorithm was proposed. Comprehensive experimentsconducted on benchmark data prove that the algorithm can replace random walkcentrality and reduce the complexity greatly.