Extracting top-k most influential nodes by activity analysis

Can we statistically compute social influence and understand quantitatively to what extent people are likely to be influenced by the opinion or the decision of their friends, friends of friends, or acquaintances? An in-depth understanding of such social influence and the diffusion process of such social influence will help us better address the question of to what extent the `word of mouth' effects will take hold on social networks. Most of the existing social influence models to define the influence diffusion are solely based on topological connectivity of social network nodes. In this paper, we presented an activity-base social influence model. Our experimental results show that activity-based social influence is more effective in understanding the viral marketing effects on social networks.