Influenced nodes discovery in temporal contact network

Abstract

Information diffusion has been studied for many years to understand how information diffuse in social network or real world. However, which nodes and when they will get influenced are unpredictable because of the uncertainty of information diffusion even we know the initial influenced nodes and diffusion network. Verification is the only way to make sure if a node is influenced or not. The target of discovering influenced nodes is to find more influenced nodes under the limited amount of verifications. In this paper, the temporal contact network is modeled. Then influenced nodes discovery problem in temporal contact network are studied based on the Independent Cascade (IC) model. A path length limited approach is proposed to calculate the infection probability approximately. Experimental results on real and synthetic data sets show our approach has better performance than BFS and Random Walk algorithm.