Degree Based Seed Optimization to Maximize Information Dissemination in Social Networks

We study the problem of optimal seed selection to maximize the fraction of individuals which has received a message in a social network. We have used the Susceptible-Infected (SI) process to model information epidemics. We formulate an optimization problem under a fixed budget constraint on the resource available to recruit individuals in the network to act as seeds, to achieve the above objective. The seeds are decided based on node degrees. This approach will work even when the exact adjacency matrix of the network is unknown and only degrees of the individuals in the network have been estimated. We study effect of the degree distribution of the network on the optimal seed selection strategy and present results for synthetic scale free and Erdos-Renyi networks, and a real scientific collaboration social network. The optimal strategy is compared with two heuristic strategies that (i) selects seeds uniformly among all degrees and (ii) selects highest degree nodes as seeds. Our results show that for a wide range of model parameters, targeting only the highest degree nodes is not optimal for various networks. This work may be of interest to advertisers and campaigners who are interested in spreading a message in a population connected via social networks