Identifying High betweenness Centrality Vertices in Large Noisy Networks

Abstract

Most real-world network models inherently include some degree of noise due to the approximations involved in measuring real-world data. My thesis focuses on studying how these approximations affect the stability of the networks. In this paper, we focus on the stability of betweenness centrality (BC), a metric used to measure the importance of the vertices in the network. We present our results on how the ranking of the vertices change as the networks are perturbed and introduce a group testing algorithm that we developed that can correctly identify the high valued BC vertices of stable networks in lower time than the traditional approaches.