Seed set selection in evolving social networks

With the popularity of social network sites, a huge volume of data has been produced. Motivated by the value of big data, a number of organizations and business companies have been putting effort into analyzing big data to increase business performance. Then many problems have been raised about social network analysis views big data, and one of key research issues is social influence analysis. Unfortunately, in this area, a large amount of research works only focused on analyzing the static social network. In detail, in this kind of network, we can use many existing efficient algorithms to select the most influential nodes which can maximize influence spread. However, the real-life network structure will change constantly over time, so a series of network snapshots will be generated. In this case, those nodes can not maximize influence spread in the following time stamps. In this paper, we propose a novel and efficient method to select seed set in evolving social network under the information of this network is only available at first time stamp. Then this seed set always makes good performance of influence diffusion during the evolution of social network. Experimental evaluations using both real and synthetic large dynamic networks show that our proposed algorithm achieves a better performance than the methods ignoring real-life social network evolution.