From complex network to skeleton: M j -Modified topology potential for node importance identification

Abstract

Node importance identification is a crucial content in studying the substantial information and the inherent behaviors of complex network. On the basis of topological characteristics of nodes in complex network, we introduce the idea of topology potential from data field theory to capture the important nodes and view it as the skeleton nodes. Inspired by an assumption that different mass of node (m j parameter) reflects different quality and interaction reliability over the network space. We propose TP-KS method that is an improved topology potential algorithm whose m j is identified by k-shell centrality. The important nodes identified by TP-KS is ranked and verified by SIR epidemic spreading model. Through the theoretical and experimental analysis, it is proved that TP-KS can effectively extract the importance of nodes in complex network. The better results from TP-KS are also confirmed in both real-world networks and artificial random scale-free networks.