On Complex Network Characteristics In Clustering Algorithm Research

Complex networks as an emerging discipline direction, greatly attractedextensive attention from researchers of different disciplines, has become a hot topic ofacademic research. People live in a world filled with a variety of complex networks.So it is necessary to study the complex network. in one word,complex networksshowing a high degree of complexity of the network, So it makes the study ofcomplex networks is very difficult. Complex network with the rapid development ofnetwork size and computer technology division multifarious methods of communitydetecting.The cluster analysis (Clustering, Analysis) is one of the main functions and tasksin data mining (Data Mining),and it has become a very active research area in datamining and exploratory analysis. To face with massive amounts of data, First andforemost task is to classify them. Cluster analysis is a reasonable classification of theraw data. However, due to the complexity of clustering objects in high dimensionalfeature space distribution Uncertainty and flexibility in the evaluation of clusteringresults and clustering as an optimization problem to solve high computationalcomplexity, The clustering algorithm is still faced with many problems andchallenges.The main work of this paper is as follows:(1)Main content of this thesis is overview of complex network and the researchsignificance, and introduce the data mining and clustering, and cluster analysis andrelated algorithms.(2)The topic focuses on the hierarchical clustering algorithm thinking andanalysis of its problems. As the nodes of the cluster center is not only a strongconnection strength have other nodes in the complex network, connected between thenodes with the interconnect density and strength, which has a strong local aggregation.This article draw on the complex network of nodes in the the eigenvalue CFidefinition, put forward the idea of the improved algorithm. Based on the introducetionthe Chameleon algorithm,combined with the definition of the Chameleon algorithm isproposed based on eigenvalue and relevantly experiments and effects analysis in the standard data set for experiments and effects analysis.(3)This thesis briefly introduces research status and related algorithms incomplex networks and improve the CNMC algorithm is applied to the complexnetwork of associations found.