The Visualization Display Of Complex Network

Complex system in real society is either existing in the form of complex network, or could be transformed into complex network by some means. However, the traditional way to organize data in the form of figures and tables makes it very hard to have a whole understanding of complex network. While the complex network visualization technology can effectively show the structure of network, making it possible to obtain useful information and timely make use of it. Due to the rapid development of information society, there generated increasing amount of data. The scale of the complex network is also growing up so that the visualization of network is becoming increasingly difficult in strict accordance with the layout algorithm. On the one hand, the limitation of computer performance causes the ineffectively performance of layout algorithm when dealing with the large amounts of data, resulting in a large number of nodes and edges overlap. On the other hand, if it is put too much nodes on the resulting pictures, that will seriously affect us to observe the network, also make it impossible to get any useful information from it. Therefore, the visualization compression algorithm comes into being.The visualization compression algorithm aims to delete operation of the nodes and links in the original network appropriately, to show the network’s topology better, to help us to have a better understanding of the complex network’s topology, to unearth the useful information of the network. It is well known that most networks’ degree distributions meet the characteristics of power law in real society. That is to say, the vast majority of the nodes’ degree is relatively small, while only a few nodes have large degree. And most of the nodes with smaller degree are less important, because its existence has a little influence on the overall topology of network. Therefore, the compression algorithm is built for the deletion of the less important nodes.This paper designs a compression algorithm on complex network, the compression algorithm mainly consists of two operations: nodes merging and node deleting. Node deleting is based on the importance of nodes, which we represent it with PageRank value in this paper. It is common to have community structure for a complex network, which means node’s importance is relative. Therefore, the compression algorithm use the RAK community mining algorithm based on label propagation to detect the community first, and then to delete the node within the community. When deleting the node, considering the connectivity of the original network, we need to add some links between nodes appropriately. Due to network scale is huge, in order to meet the requirements of accuracy and efficiency, we make use of the GraphLab parallel framework.This paper implements the complex network compression algorithm, and the PageRank calculation and the community detection are implemented under the GraphLab parallel framework. We validate the accuracy and efficiency of PageRank algorithm and RAK algorithm implemented under the GraphLab framework, and tests the compression algorithm both on the actual data set. We find that the compressed network still meet the power law distribution, which verifies the validity of the compression algorithm. The results of the compression algorithm is shown at the end of this paper.