Integrated evaluation approach for node importance of complex networks based on relative entropy

Evaluating the node importance correctly is a crucial issue for complex networks research. If a network has multiple or ambiguous morphological features, the evaluation result of an individual index may be unilateral. Meanwhile, the results may be inconsistent if different indexes are adopted simultaneously. To solve the problem, an integrated approach is proposed based on relative entropy. In this approach, a system of multiple indexes is constructed firstly. Then each evaluation result of an individual index is handled into a discrete distribution. Finally an optimal integrated evaluation solution is obtained by linear programming. This approach has a well-formed theoretic basis and an easily calculated procedure, which can be used in a variety of complex networks. Experimental results show that the proposed approach is more effective than other different methods proposed in some literatures.