Weighted Complex Network Key Node Identification Technology Based On Link Relations

As the hottest topic in discipline field, the study of Complex Network is receiving more and more attention from researchers, Its basic theory has penetrated into many subjects and fields. As an important branch of which, Mining vital nodes has profound theoretical significance and practical value. Throughout the achievement, there still are two problems to be solved:1) The evaluation methods existing are most to undirected and unweighted network, However, lots of real-world networks that can be modeled as Complex Networks are Directed Weighted Network, So it need to find new evaluation metric and method; 2) It's lack of correlation analysis In comprehensive application of evaluation metrics of vital nodes, Most methods just use the metric signally or one after another, which can be result in losing accuracy of evaluation result.In order to improve the practicality and accuracy of vital nodes recognition, after analys-ing the structure characteristics of Directed Weighted Network, This paper promoted a new evaluation metric and new evaluation method as follows:1) Based on the structure characteristics of Directed Weighted Network, by delving into the Directed Weighted Network Model which based on BBV building thought, certain the related parameters and design experimental to simulate the model, prove that the node in the model has law properties.2) Select evaluation metric. This paper puts forward a new concept named as DWNodeRank, Which as a new evaluation metric for node importance. This metric based on the link structure, especially direction and weights, takes into account both the local statistical characteristics and entire link characteristic of Directed Weighted Network.3) Based on the relevance between metrics, use appropriate center metric to define initial vector, that in order to speed up the convergence of the assessment. Then promote a new evaluation algorithms according to DWNodeRank value and related statistical characteristics of nodes, this algorithms has a time complexity of O(cn2).Finally, through the simulation experiment shows the proposed algorithm is effective to evaluate the node importance in Directed Weighted Network. It proves that this method has certain advantages from aspects of algorithm thought and computational complexity.