An Effective Measurement Scheme of Node Influence in Aviation Network

Aviation network is the typical information network, where the airports can be regarded as nodes, and the airlines between airports can be regarded as links. But with the constraints with the Degree-Centrality, Closeness-Centrality and the Centrality between nodes, traditional researches on the influence of the information network developed slowly. Thus, we do some experiments on the traditional methods which used in the researching on the influence of the information network. And during experiments, we find that traditional methods have two drawbacks: (1) Universality is an unpredictable factors because some indicators are not suitable for any occasion. For instance, when the size of information network extends very large, it may needs a lot of time to compute the Closeness-Centrality and the the Centrality between nodes. (2) Node is such difficult to select correctly. For example, it is an NP-hard problem to select K as the most influential node because it has a very high computational complexity. This paper proposed a new perspective to depict the node influence in information network: By doing research on susceptible sides of the node depicts the node influence in aviation network. Simultaneously, this scheme includes the measurement method of non-directional and directional network. And we use the Independent-Cascade-Model to evaluate the feasibility and effectiveness of our method. Because of the lack of real aviation network data, we apply our method on three similar real network data-sets (Hamsterster full, Ca-HepTh etc). And the result refers to the experiments show that our method exceeds the greedy algorithm of influence maximization on the spread influence. Moreover, this also show that our method not only can be suited to the aviation networks, but also can be used in other information networks. Therefore, Our scheme can accurately depict the node influence in information network.