Improving information centrality of a node in complex networks by adding edges
The problem of increasing the centrality of a network node arises in many practical applications. In this paper, we study the optimization problem of maximizing the information centrality I_v of a given node v in a network with n nodes and m edges, by creating k new edges incident to v. Since I_v is the reciprocal of the sum of resistance distance R_v between v and all nodes, we alternatively consider the problem of minimizing R_v by adding k new edges linked to v. We show that the objective function is monotone and supermodular. We provide a simple greedy algorithm with an approximation factor (1-1/e) and O(n^3) running time. To speed up the computation, we also present an algorithm to compute (1-1/e-ϵ)-approximate resistance distance R_v after iteratively adding k edges, the running time of which is O (mkϵ^-2) for any ϵ>0, where the O (·) notation suppresses the poly ( n) factors. We experimentally demonstrate the effectiveness and efficiency of our proposed algorithms.
READ FULL TEXT