The Research And Application Of Community Detection By Using Clustering Algorithm

Complex network is a research hotspot in recent years,people also pay attention to the community structure besides the small world and scale-free network features.The nodes within the same community connect closely,while between the communities connect loosely.Based on the characteristics of community structure,this paper has proposed the CDNIM algorithm.Through validly mining the community existed in complex networks,the topological structure and function of complex networks can be analyzed,and hidden rules can be found as well.Based on the elicitation of spectral bisection method,node important matrix is defined in this paper,and combined with the clustering algorithm,an community partition algorithm(A Community Discovery Algorithm based on Node Importance Matrix,simply called CDNIM)based on node importance is proposed.This method combines the characteristics of node important degree matrix and spectral bisection method,the data are processed and use K-means clustering algorithm to classify the nodes in the network.To find the optimum number of community network search,the module function is introduced.When the module value is the largest,the community structure is corresponding to the best division of the network community.This algorithm CDNIM is applied in many kinds of classical data sets--the karate club network,dolphins networks,football network,etc and validated its accuracy and efficiency.The experiments also indicate that the CDNIM algorithm is better than those classical algorithms such as K-L,GN,etc on classification of the community's accuracy rate.