Fast ranking nodes importance in complex networks based on LS-SVM method

Abstract

Achieving high accuracy and comprehensiveness in node importance evaluation of complex networks is time-consuming. To solve this problem, a method based on Least Square Support Vector Machine (LS-SVM) was proposed. Firstly, four complicated importance indicators which reflect the node importance globally and comprehensively were selected. Then analytic hierarchy process (AHP) method was applied to obtain the node’s importance evaluation. On this basis, three simple indicators with low computational complexity were proposed, and LS-SVM was adopted to find the mapping rules between simple indicators and AHP evaluation. The experiments on artificial network and actual network show the validity of proposed method: the evaluation based on complicated indicators is consistent with reality and reflects node importance accurately; simple indicators evaluation by LS-SVM saved a lot of computational time and improved the evaluating efficiency. Our method can provide guidance on influential node identification in large scale complex networks.