Finding top-k influential users in social networks under the structural diversity model

The influence maximization problem in a large-scale social network is to identify a few influential users such that their influence on the other users in the network is maximized, under a given influence propagation model. One common assumption adopted by two popular influence propagation models is that a user is more likely to be influenced if more his/her friends have already been influenced. This assumption recently however was challenged to be over simplified and inaccurate, as influence propagation process typically is much more complex than that, and the social decision of a user depends more subtly on the network structure, rather than how many his/her influenced friends. Instead, it has been shown that a user is very likely to be influenced by structural diversities of his/her friends. In this paper, we first formulate a novel influence maximization problem under this new structural diversity model. We then propose a constant approximation algorithm for the problem. We finally evaluate the effectiveness of the proposed algorithm by extensive experimental simulations, using different real datasets. Experimental results show that the users identified from a social network by the proposed algorithm have much larger influence than that found by existing algorithms.