Taking the Foursquare data as an example, this paper investigates the problem of finding influential nodes in a location-based social network (LBSN). In Foursquare, people can share the location they visited and their opinions to others via the actions of checking in and writing tips. These check-ins and tips are likely to influence others on visiting the same places. To study the influence behavior in LBSNs, we first propose the attractiveness model to compute the influence probability among users. Then, we design a one-wave diffusion model, where we focus on the direct impact of the initially selected individuals on their first degree neighbors. Base on these two models, we propose algorithms to select the k influential nodes that maximize the influence spread in the complete-graph network and the network where only the links with friendship are preserved. We empirically show that the k influential nodes selected by our proposed methods have higher influence spread when compared to other methods.
展开▼
