Applying an Influence Measurement Framework to Large Social Network

Predicting influential users is one of the important research problems on social network analysis. It helps to understand many complicated phenomena including information dissemination. It can be employed in many real world applications such as viral marketing. Influential users can influence social network users using their attributes, strategic locations or expertises. In this paper, we tackle this problem by proposing a novel hybrid framework that is used to predict influential users on social network. In this framework, we integrate users’ attributes and their strategic location to measure influence. We apply several centrality analysis algorithms to find user’s strategic locations, while we adapt a real world attribute measure that is used by Flickr based on users’ attributes. We employ our framework to a large dataset crawled from real world social network, i.e., Digg. We evaluate the proposed framework in term of correlation. We further show that the proposed framework outperforms other measurements