Research On Identifying Key Nodes Based On Heterogeneous Networks

Among the fundamental research on the complex network, the study on the key nodes is a hot filed. Although take a small proportion in number, the key nodes often play important roles in the whole network and determine the network function.Thus studying the vital nodes in a network is very valuable. Recently there are many kinds of methods on identifying the key nodes, but due to the complexity factors of the network structure, function as well as the specific needs of practical applications, these methods can’t meet the need. Therefore, how to evaluate with high efficiency the importance of the nodes and find the key nodes matching with the actual situation have become the research hotspots.To address these problems, we do some work and improvement as follows:1. We introduce the way how to store a complex network in the computer and the common statistical feature, we also give the corresponding calculation formula. Meanwhile, in this paper we introduce four basic models of complex networks in detail, construct and display in the form of graph respectively.2. We establish the heterogeneous network model to represent the real world more accurately in some extend and analysis the characteristics. Then we use five classical methods to identify the key nodes on the heterogeneous network and find out that the importance contribution matrix method and the K-shell method have a better recognition effect than the other three methods.3. We put forward the concept of "key node" and "importance" to help to find out the target nodes explicitly. In the view of the node relative importance evaluation, we propose the assessment framework and give three evaluation strategies. Through the study on the reliability of networks, we use statistical characteristics evaluation method to formulate and quote the reliability of the network connectivity and network vulnerability index, implement the evaluation on the node relative importance and the discovery of the key nodes. The feasibility and effectiveness of the method is also verified through experiments. This method overcomes the defects of traditional single network performance evaluation, gives a comprehensive and effective measurement for the network reliability. Also, this method shows the whole process of finding the key nodes objectively and provides reference instance for the future study about identifying the key nodes in the heterogeneous network.During the research on the techniques of key nodes discovery, we take full account of the influence that the importance of network structure effect on the nodes evaluation, through the analysis and solution to the relative importance of nodes, figures out the problem of the discrepancy in traditional nodes evaluation. Meanwhile, it consummates those methods of nodes weightiness evaluation in existence.Finally, we summarize the whole work on the complex network nodes discovery, point out the shortages in the research process and give the outlook for the research in future.