The Technology Research Of The Optimize Algorithm Of Community Division On Random Walk

Complex network is widely existed in nature and human society. For example, human social network, biological network, Electronic circuit network and so on. In order to further study the meaning and value of network, it is necessary to analyze the network structure systematically. Currently, the study on complex network has penetrated into all fields, and the algorithm of community division is the significance core in the research area of the complex network.With the development of social networks and web technology, the size of the network has increases gradually, which requires efficient and fast community division algorithm on the large complex network, and has been a hot area. To make it, this paper obtains the inspiration from the PageRank algorithm, using the random walk. Let each node in the network has the energy, the energy of the nodes transfer their energy by the energy transition probability matrix, when the node will reach a steady state after energy transfer, the energy matrix is obtained. The two nodes are divided into a same community by analyzing the energy matrix,which transfer more energy with each other than the other nodes. Gradually implement the above principle of community division, until the whole network is divided into a community,and the optimal results are achieved by the Q value. The core idea of the algorithm is said above.With the increase of network scale, the algorithm needs a lot of storage space. As the computer can't supply the enough storage space, effective results are unable to get. It's widely known that social network node degrees obey power-law distribution, according to the predecessors' research. So the adjacency matrix and the energy matrix of the large complex network community are the sparse matrix. In order to further optimize algorithm, the sparse matrix compression storage is used to reduce algorithm need storage space in the implementation process. However, the matrix will gradually become dense matrix during the process of energy transfer. In the paper, the sparse nature of matrix sustained by settingthreshold according to the dimensions of matrix, the necessity of setting threshold value and how to set the threshold to maintain the sparse matrix are illustrated by experiments, the computation efficiency of the algorithm is improved. To verify the accuracy of the algorithm proposed in this paper, through the experiments on classic small sets is known accurate results,the accurate result is received. The good community division results and higher efficient calculation are acquired by the verification of large-scale network community sets.