Skip to Main Content
It is important to identify node role and track node evolution in temporal networks in many applications. Most existing methods identify the role of a node according to its static structural property. In this paper, we propose a new representation named quantitative temporal directed graph to represent temporal networks, which differs from other network representations in that it adds quantitative attributes to nodes and edges at different time points. We identify node role and track node evolution by analyzing the temporal behavioral characteristics of nodes. The distributions of nodes in different roles on the amplitude-time feature space are determined by support vector machine. We perform experiments on a dataset extracted from a large scale bulletin board system. The experimental results demonstrate the utility and distinctiveness of our method.