Efficient selection of influential nodes for viral marketing in social networks

Abstract

There has been an exponential growth in social networking and online shopping with the internet revolution in the recent past. Viral marketing, exploiting social networks to promote various products, has proved successful in influencing the public compared to other media. One well known task in this area is to choose the best nodes that maximize the overall influence propagated in the social network. t-Influence Maximization problem has been addressed in this paper which can be defined as maximizing the overall influence in a social network by selecting seeds for the given t products with their seed requirements. Constraints are set on number of products to be recommended for any single user (to avoid spamming) and total number of seeds to be selected for a particular product (budget constraint). The technique we have used, a greedy algorithm to the aforementioned t-influence maximization problem. It not only allocates seeds with maximum total influence, but also ensures that any one product doesn't dictate the overall influence and a fair selection is done. An efficient algorithm is designed for calculating the approximate influence of the selected nodes which is important in practical situations. Effectiveness and scalability of the algorithm is analyzed and verified using simulations on real-life facebook data.