A two-stage algorithm to estimate the source of information diffusion in social media networks

We study the important problem of source localization in the context of information spreads in large social networks. Specifically, we design a Maximum-Likelihood source localization algorithm that is especially suited to large social networks. Our proposed algorithm requires about 3% fewer sensor nodes than other single stage algorithms for the same level of accuracy in detection. For practical social networks, which are typically large, this translates to a significantly fewer number of sensor nodes.