We consider the problem of \emph{influence maximization}, the problem of
maximizing the number of people that become aware of a product by finding the
`best' set of `seed' users to expose the product to. Most prior work on this
topic assumes that we know the probability of each user in