A Hybrid Framework to Predict Influential Users on Social Networks

Predicting influential users is one of the major research topics in social network analysis. It can be used in many applications including marketing, recommendation systems and search engines. Influence can be shown by users' attributes, strategic locations, and expertises. In this paper, we integrate both users' location in a network and attributes to quantify their influence. In order to improve the performance of influence measurement, we propose a hybrid framework to predict influential users on social networks. The users' locations can be computed using centrality analysis algorithms, while users' attributes are users' characteristics on social networks such as activeness. We employ our hybrid framework, location-based influence measurements and attributed-based influence measurements to Flickr. The experimental results show that the proposed framework outperforms other measurements in term of correlation.